Search

Copper‐gold magmatic‐hydrothermal systems dominate in the Chibougamau area of the Neoarchean Abitibi subprovince (greenstone belt) of the Superior Province (craton), whereas orogenic gold mineralization is more common in the rest of the Abitibi. Understanding differences in metal endowment within the Abitibi greenstone belt requires insights into the geodynamic evolution of the Chibougamau area. This was addressed by imaging the crust using seismic reflection profiling acquired as part of the Metal Earth project. Seismic reflection sections display shallowly south‐dipping reflectors located within the upper crust (e.g., a possible deep continuation of the Barlow fault) and a northward‐dipping midcrust imbricated with older crust (Opatica subprovince) to the north. Multiple reflectors characterize the upper part of the midcrust, interpreted as faults superimposed on a major lithological boundary. These structures likely formed during terrane accretion prior to craton stabilization. Combining the new seismic data with known stratigraphic, structural, and magmatic records, we propose that the study area was initially a normal (i.e., thick) Archean oceanic crust that formed at or before 2.80 Ga and that evolved through terrane imbrication at 2.73–2.70 Ga. Shortening caused rapid burial, devolatilization, and partial melting of hydrated mafic rocks to produce tonalite magmas that may have mixed with mantle‐derived melts to produce the diorite‐tonalite suite associated with observed Cu‐Au magmatic‐hydrothermal mineralization.

International audience; Investigation of gold metallogeny in the Paleozoic Meguma terrane (Canada) is conducted through LA-ICP-MS analysis of arsenopyrite collected from eight slate-belt style vein gold deposits using a novel approach integrating elemental distribution maps and their derived elemental paragenesis with multi-element binary plots. The data reveal two distinct gold events: 1) an early event characterized by a Co-Ni-Mo-Sb-Se elemental association related to initial growth of arsenopyrite that reflects the presence of invisible gold (>10 ppm); and 2) a second event, spatially associated with late fracture sets, that is characterized by an Al-Ti-V-Mn element association and reflects either remobilization or upgrading of primary invisible gold and is manifest as visible gold. The results of this study indicate a complex and protracted history of gold mineralization which has important ramifications for the Meguma gold deposits, as well as other orogenic gold districts globally. In the case of the Meguma Terrane, it involves an initial gold event that is followed by element mobilization and, in the case of precious metals (Au, Ag), an upgrading through a zone refining process. In addition, the variable coupling and decoupling of elements is only revealed using in-situ derived LA-ICP-MS data.

New field observations from the Pontiac subprovince, Superior Province, supported by whole-rock geochemistry provide strong evidence for extensional magmatism and volcanism during deposition of the ca. 2685 Ma Pontiac sediments. Contacts between basaltic and komatiitic volcanic rocks and sedimentary rocks are concordant and weakly deformed, inconsistent with an allochthonous emplacement model proposed for volcanic successions in the Pontiac subprovince. Structural observations and aeromagnetic imagery show that the volcanic rocks are present as thin ( 30 km) conformable successions within the turbiditic succession and were emplaced prior to the earliest deformation of the Pontiac subprovince. Sediment-matrix breccias with fluidal and blocky igneous clasts identical to the compositions of the mafic–ultramafic rocks occur along intrusive contacts with Pontiac Group wacke. These textures are interpreted as peperite, representing the interaction of magma or lava with wet, unconsolidated sediment. Peperite was observed along the margins of tholeiitic-komatiitic rocks and calc-alkaline lamprophyre intrusions, indicating that these magmas were broadly synchronous with the deposition of the Pontiac sediments. The volcanic rocks are characterized as tholeiitic basalt and Al-depleted to Al-undepleted komatiite, have high Nb/Th, flat to positive light rare earth element (REE) slopes, and flat heavy REE slopes, consistent with partial melting of a mantle plume. The calc-alkaline lamprophyre intrusions have low Nb/Th and prominent negative light and heavy REE slopes, consistent with partial melting of a previously metasomatised mantle. The presence of autochthonous komatiitic and tholeiitic rocks is consistent with the Pontiac subprovince developing as an extensional basin within the Abitibi subprovince in response to an upwelling mantle plume ca. 2.7 Ga. This model is incompatible with uniformitarian evolutionary models for the Superior craton interpreting the sedimentary subprovinces as accretionary prisms that formed during subduction processes. Non-uniformitarian models invoking stagnant-lid tectonics predict the formation of extensional basins and ascent of depleted mantle-derived magmas associated with upwelling mantle plumes, but the ability to produce source-metasomatism magmas without subduction is highly contested. The evolution of the Pontiac basin may thus reflect a transitional period in the Neoarchean where stagnant-lid and proto-subduction styles of tectonism were operating intermittently.

Algoma-type banded iron formations (BIF) are chemical sedimentary rocks characterized by alternating layers of iron-rich minerals and chert that are generally interstratified with bimodal submarine volcanic rocks and/or sedimentary sequences in Archean greenstone belts. However, the geological setting for Algoma-type BIF deposition remains equivocal due to the effects of post-depositional deformation and metamorphism, and absence of modern analogues for comparative studies. It is commonly accepted that the abundance of rare earth element and yttrium (REE + Y) in chert bands may retain a primary geochemical signature and therefore constrain their geological setting. In order to explore the latter, a geochemical study using the laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) methodology was done using cherts from four Canadian BIF-hosted gold deposits. These results suggest that chert bands record: (1) interaction of seawater with Fe-oxyhydroxides, as suggested by their heavy REE enrichment coupled with La and Y enrichments; (2) contributions from high-temperature (>250 °C) hydrothermal fluids, as suggested by positive Eu excursions; and (3) detrital contamination, which is suggested by relatively consistent REE concentrations and a chondritic Y/Ho ratio (i.e., Y/Ho ≈ 27). Water-column pH conditions at the time of BIF deposition are evaluated using Ce/Ce∗: a positive Ce/Ce∗ anomaly suggests relatively acidic conditions (i.e., pH ⩽ 5) for most of the chert samples, but more alkaline conditions (i.e., pH ⩾ 5) for samples showing Fe-oxyhydroxide precipitation within chert bands. Finally, in situ using secondary ion mass spectrometry (SIMS) analysis (n = 73) of chert from Meliadine show the δ18O of primary amorphous silica (+27‰) was modified to values of around +8‰ to +20‰ during diagenesis at temperatures >100 °C with a fluid having δ 18 O H 2 O = 0 – 5 ‰ . Thus, whereas there has been O isotopic exchange during diagenesis, the REEs and trace elements are not modified in the chert due to the low concentrations of these elements in the reacting fluid of sea water origin.

International audience; Orogenic gold deposits span a spectrum in regards to setting and style of mineralization, nature of gold (e.g., invisible, coarse) and favourable host rock. In addition, deposit formation is often attributed to protracted multi-stage hydrothermal processes. That gold mineralization also depends on a variety of features, such as metal-source reservoirs, metal-transport processes and wall-rock stratigraphy, which add to the challenge of investigating origin of paleo-mineralized settings. Recent work using quantitative laser ablation inductively coupled plasma-mass spectrometry (LA ICP-MS) element distribution maps/profiles and their corresponding time slice datasets (TSD) provides new insight into identifying and assessing elemental paragenesis, multi-dimensional element coupling/decoupling processes, and corresponding mineralizing events. To further assess complexities of mineralization, application of geostatistical tools (e.g., multidimensional scaling, principal component analysis, linear discriminant analysis) and various innovative multi-element binary plots (e.g., Ag versus Au, Ni versus Co) is advised. To illustrate the application of this methodology, the results LA ICP-MS mapping and data processing for Fe sulphides from several metasedimentary-rock-hosted Canadian gold systems are presented: three Archean Algoma-type BIF-hosted gold deposits (~ 4 Moz Au Meadowbank, ≥ 2.8 Moz Au Meliadine district, ~ 6 Moz Au Musselwhite) and eight slate-belt style vein gold deposits from the Paleozoic Meguma terrane (Nova Scotia). The maps and derived elemental plots generated from the various settings demonstrate that: 1) the gold mineralization present is the product of multi-stage processes; 2) elemental associations vary as mineralization progresses, such as the early growth history of sulphides versus later coupled dissolution-precipitation reactions; and 3) different metal-source reservoirs and stratigraphy influence the fluid signature. These results contribute to better deciphering the complex processes involved in the protracted evolution of these and other orogenic-type gold systems (e.g., remobilization of invisible gold from early sulphide, precipitation of visible gold in later sites, increase of Au fineness).

International audience; Among the many types of mineral deposits within Archean cratons, gold mineralization is an important economic commodity with over 20,000 metric tons of gold produced from greenstone belts in 2001. Of the Archean-early Paleoproterozoic gold deposits, several different types of mineralization are known, which includes the important Algoma-type banded iron formation (BIF) where gold is locally associated with sulfide-facies zones within regionally extensive oxide-facies. It is commonly accepted that the shale-normalized chemical signature of REE + Y of chert bands in Algoma-type BIFs may reflect one of the three processes, each of which may be relevant to the nature and origin of the gold mineralization: (1) direct seawater precipitation; (2) involvement of and contribution from hydrothermal fluids; and (3) replacement of precursor volcanic units due to silicification. An essential question in regard to the mineralization is, therefore, whether the gold mineralizing fluids have a preference for one geochemical type of iron formation versus another. In order to assess the relevance of these competing models, we report herein the results of a LA ICP-MS study of chert samples within different Algoma-type BIFs from the Meadowbank deposit (24.5 Mt proven/probable ore reserves grading 2.8 g/t (2011)) hosted in the Neoarchean Woodburn Lake Group of the Rae Domain of the western Churchill Province, Canada. This study used 39 carefully selected and characterized (i.e., petrography and SEM-EDS imaging) chert samples from the main deposit, the Central BIF, and four additional BIFs, the Far West, West, East and Grizzly zones, with data collected using line traverses along the chert bands. The geochemical data indicate that an ambient seawater signature (i.e., enrichment in HREE relative to LREE associated with positive La and Y anomalies) dominates the samples with a lesser hydrothermal component (characterized by a positive Eu anomaly) and the influence of detrital contamination can also be detected. These initial results indicate that the methodology and protocol employed provides a reliable means to assess and interpret the chemical signature of BIFs hosting gold mineralization. In the present case, the results for the Meadowbank deposit suggest that chert from mineralized BIF units does not record a typical chemical signature that may be used as a vector for potential gold mineralization.

International audience; Quantitative laser ablation inductively coupled plasma-mass spectrometry (LA-ICP-MS) element distribution maps combined with traverse mode analyses have been acquired on various sulfides (pyrite, pyrrhotite, arsenopyrite) from three Canadian Algoma-type BIF-hosted gold deposits (~4 Moz Au Meadowbank, ≥ 2.8 Moz Au Meliadine district,~6 Moz Au Musselwhite). These data, in conjunction with detailed petrographic and SEM-EDS observations, provide insight into the nature and relative timing of gold events, the presence and implication of trace element zoning regarding crystallization processes, and elemental associations that fingerprint gold events. Furthermore, the use of an innovative method of processing the LA-ICP-MS data in map and traverse modes, whereby the results are fragmented into time-slice data, to generate various binary plots (Ag versus Ni) provides a means to identify elemental associations (Te, Bi) not otherwise apparent. This integrated means of treating geochemical data, along with petrography, allows multiple gold events and remobilization processes to be recognized and their elemental associations determined. The main gold event in each of these deposits is characterized by the coupling of an As-Se-Te-Ag element association coincident with intense stratabound sulfide-replacement of the Fe-rich host rock. Additionally, the data indicate presence of a later remobilization event, which upgraded the Au tenor, as either non-refractory or refractory type, along fracture networks due to the ingress of subsequent base metal-bearing metamorphic fluids (mainly a Pb-Bi association). Furthermore, the data reveal a stratigraphic influence, as reflected in the elemental associations and the elemental enrichments observed and the nature of the sulfide phase hosting the gold mineralization (arsenopyrite versus pyrite).

International audience; The geochemistry of chert layers in Algoma-type banded iron formations (BIF) has been used to constrain the depositional setting of the BIFs, as rare earth element (REE) and yttrium (Y) systematics are a function of their chemical environment of formation. In contrast, the chemistry of the interbedded oxide-rich layers (i.e., magnetite) has not been analyzed for this purpose because of the presumed potential effects related to diagenetic changes during conversion from primary iron-bearing mineralogy to magnetite. Here, we explore the validity of this latter hypothesis by examining the results of LA-ICP-MS analysis of iron-oxide layers at three Canadian BIF-hosted gold deposits (i.e., Meadowbank, Meliadine, Musselwhite) to assess whether the primary REEþY systematics of the oxide layers are preserved and, if so, what are the implications. The results indicate that, regardless of their diagenetic, later metamorphic, and hydrothermal histories, the chemistry of the iron oxides retains a primary signal in all cases with the following indicated: (1) interaction of the primary Fe-oxyhydroxide phases with seawater, as reflected by heavy REE enrichment relative to light REE depletion that is coupled with variable La and Y enrichment; and (2) some input of moderate-temperature (.250 8C) hydrothermal vent fluids, as suggested by positive Eu anomalies. The chemical data are also used to evaluate currently used classification diagrams for ore deposits based on magnetite chemistry. Our new data indicate that the fields for magnetite geochemistry in BIF are too restricted or lead to misclassification of samples. In the case of the latter, it may be that interaction of fluids with the immediate substrate influences the chemical signature of samples. Therefore, caution is suggested in using these diagrams where hydrothermal fluids are involved in magnetite formation.

This study reports new REE+Y and 3-isotope sulfur data for Archean banded iron formation and volcaniclastic rocks with cherty bed tops in the Neoarchean Abitibi greenstone belt of Canada. The data were analyzed with a view to better constrain Neoarchean ocean chemistry, atmospheric conditions prevalent during weathering and transport, the development of Algoma-type banded iron formation and the overall process of stratigraphic development of greenstone belts. The Abitibi greenstone belt consists of 7 mafic to felsic volcanic cycles, each capped by a sedimentary interface zone consisting of chemical and minor clastic metasediments. We concentrated sampling on the iron formation capping the ca. 2730 Ma Deloro assemblage as it occurs over a wide area (300 km x 600 km) and because there is a substantial depositional gap prior to deposition of the overlying volcanic rocks. Volcaniclastic rocks within the ca. 2710 Ma Tisdale assemblage were also sampled. Chemical analyses focussed on the SiO2-rich portion of the samples and were conducted by laser ablation ICP-MS. In situ analysis of S isotopes was obtained for pyrite by ion probe. REE data display four types of patterns: (1) hydrothermally influenced marine hydrogenous sediment, (2) contaminated, hydrothermally influenced marine hydrogenous sediment, (3) hydrothermally dominated patterns, and (4) replacement patterns indicating silicification of precursor volcanic units. Contamination and/or the presence of non-chert components were documented with Th. U and Zr content. Non-chert components were defined as: (1) phosphates that led to elevated Th/U, (2) clastic detritus leading to flat shale normalized REE patterns, and (3) volcanic detritus leading to elevated values for Zr. No meaningful difference in REE+Y geochemistry as a function of elevated Th/U was found implying that phosphates have the same REE patterns as the host chert. The cherts within banded iron formation exhibited stratigraphic variation in several localities, progressing from replacement chemistry (flat REE profile) at the base, hydrogenous sediment geochemistry (positive La, Gd, Y/Ho anomalies) in the middle part and hydrothermal patterns (depleted LREE, elevated positive Eu anomalies) in the upper part. The upper parts of some units also display +Ce anomalies possibly reflective of more oxygenated water also supported by S isotope data. A consistent increase in Pr/Vb in the upper parts of units is postulated to reflect shallowing upward of depositional depth to an unknown extent but not above storm wave base. A number of samples with flat REE patterns lacking La and Gd anomalies represent hydrothermal deposition with the largest Eu/Eu* values recorded for Archean iron formation. The main contribution of the Abitibi banded iron formations is that they provide a deeper water perspective on Archean ocean chemistry. The resulting picture is that of slow BIF accumulation, a generally strong hydrothermal input of REE and complex oceanic cycling, possibly involving a chemocline above the sampled water depth. The new sulfur isotope data show a greater extent of mass independent fractionation than previously recorded for the Neoarchean, with 6,Delta S-33 ranging from -1.4 to +4.2 parts per thousand.. The strongly MIF positive source (Delta S-33=+4 parts per thousand) was apparently similar to Paleoarchean values. (C) 2012 Elsevier B.V. All rights reserved.

article i nfo We studied heavy minerals extracted from a diamondiferous metaconglomerate that formed 2697-2701 Ma in a successor basin within the Michipicoten Greenstone Belt (MGB) of the Wawa-Abitibi Terrane (Southern Superior Craton). The conglomerate is metamorphosed in the greenschist facies and contains mainly locally derived igneous mafic to felsic detritus, but also very minor components of medium grade metamorphic minerals, diamonds and paragenetic diamond indicator minerals. Comparison of the size distribution, resorb- tion and N aggregation of diamonds in nearby Wawa lamprophyres and the metaconglomerate diamonds confirms that the latter were not derived from the proximal lamprophyric source. The heavy minerals in the metaconglomerate include diopside, olivine, corundum, chromite, almandine, pyrope with kelyphitic rims, picroilmenite, amphibole and anorthite. Low abundances of the heavy minerals (several grains per 4-70 tons of the metaconglomerate) are, in part, explained by their complete or partial replacement by the greenschist mineral assemblage. Detrital almandine and amphibole are inferred to originate in amphibolite facies rocks. Cr-diopside, olivine, chromite and anorthite were sourced from mafic-ultramafic anorthosite- and chromitite-bearing layered complexes mapped in the MGB. The presence of pyrope with more than 6 wt.% Cr2O3 suggests derivation from a cratonic root. Picroilmenite has compositions typical of kimberlite and unlike that of ultramafic lamprophyres and other unconventional diamondiferous volcanics. The Wawa metaconglomerate, therefore, should be considered analogous to the Witwatersrand successor basin conglomerate in recording indirect evidence for Archean kimberlites. The tight localization of the diamondif- erous conglomerate in time and space was controlled by a quick (~3 Ma) erosion of the source kimberlite body. The location of the kimberlite-bearing N2.7 Ga Superior protocraton was inferred from the provenance of the metaconglomerate detrital material. The clasts could have originated as close as the northern Wawa- Abitibi Terrane or as distant as the Opatica terrane. The pre-2.7 Ga diamonondiferous cratonic root below the Southern Superior was removed in the Neoarchean-Proterozoic. The existence of Archean kimberlites and deep diamondiferous roots below smaller pre-2. 7 Ga protocratons emphasizes the similarity of Neoarchean and Phanerozoic mantle processes.

The interbedded chert and iron minerals of banded iron formation (BIF) provide an important window into the chemistry of the Precambrian ocean. Chert microbands have better preservation potential than the iron-rich bands but pose analytical challenges. The generally much lower concentration of trace elements in chert require better detection limits and make chert more susceptible to the influence of foreign materials such as ash or detritus. Without screening for chemical complexity, bulk chemical analyses may have limited significance. Here we report a new analytical protocol consisting of two steps. First it uses exploratory laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) in situ analysis for identifying suitable chert areas to be sub-sampled for solution ICP-MS analysis. Second, the subsampled chert is digested and dried down, volatilizing the bulk of the matrix as SiF4. The solution can then be admitted to the ICP-MS at much lower dilution factors than common rock samples, resulting in detection limits for many of the rare earth elements at double or single digit ppt levels. This approach produces high quality trace element data even for cherts with exceedingly low trace element contents. Significantly, such data approach primary chemical composition of the BIF better than conventional bulk analysis. It is also demonstrated, through selective digestions, that the most common types of mineralogical heterogeneity in cherts - carbonate and iron-oxide minerals - have minimal effect on the REE + Y composition, at least in studied cherts from the Abitibi Greenstone Belt. The proposed combination of techniques will allow future studies of more refined trace element chert geochemistry, providing a better understanding of Precambrian oceans and BIF deposition. (C) 2011 Elsevier B.V. All rights reserved.

The Abitibi greenstone belt contains some of the world’s largest Cu-Zn and Au deposits and significant amounts of Ni-Cu-PGE mineralization. Discoveries early in the 20th century were made by classic prospecting with subsequent discoveries commonly resulting from diamond drilling of targets identified by electromagnetic surveys. The next generation of deposits will be found at greater depth, based on geological and geophysical models, principally beneath the extensive overburden that covers most of the Abitibi greenstone belt. Thus, exploration will require improved knowledge of the geologic characteristics of the existing deposits, particularly the relationship between deposit- and regional-scale features, including the stratigraphic, plutonic, structural, and metamorphic architecture of the region, and will rely more heavily on expensive geophysical and geochemical techniques, verified by diamond drilling. The papers in this Special Issue are part of the Discover Abitibi Initiative, a geoscience program funded by federal, provincial, and municipal governments in collaboration with the mining industry. Many of the results reported here are from projects funded by Phase II of the Discover Abitibi Initiative which commenced in 2003 and was focused on the Timmins-Kirkland Lake region (Fig. 1⇓), a high mineral potential portion of the Abitibi greenstone belt that has been the source of exceptional mineral wealth. As part of the Discover Abitibi Initiative, the Greenstone Architecture Project was undertaken by the Mineral Exploration Research Center at Laurentian University to bridge the gap between academic and company information at the scale of individual deposits and the broad-based knowledge derived from government mapping. The project was designed as a multidisciplinary approach involving faculty, postdoctoral fellows, and graduate students in collaboration with geoscientists from the Ontario Geological Survey, the Geological Survey of Canada, consultants, and company geologists. As can be seen in the papers in this Special Issue, the program had both geological and …

Models of greenstone belt development are crucial for exploration. Allochthonous models predict belts to be a collage of unrelated fragments, whereas autochthonous models allow for prediction of syngenetic mineral deposits within specific stratigraphic intervals. Superior province greenstone belts consist of mainly volcanic units unconformably overlain by largely sedimentary “Timiskaming-style” assemblages, and field and geochronological data indicate that the Abitibi greenstone belt developed autochthonously. We describe major revisions to stratigraphy of the Abitibi greenstone belt and the implications of an autochthonous development of the volcanic stratigraphy for exploration for syngenetic mineralization. The Abitibi greenstone belt is subdivided into seven discrete volcanic stratigraphic episodes on the basis of groupings of numerous U-Pb zircon ages of pre-2750, 2750 to 2735, 2734 to 2724, 2723 to 2720, 2719 to 2711, 2710 to 2704, and 2704 to 2695 Ma. We present revised lithotectonic and/or stratigraphic nomenclature using these time intervals, including (1) isotopic inheritance in younger episodes which indicates that the older episodes (2750–2735 and 2734–2724 Ma) had greater extent than is presently seen, (2) dikes feeding younger volcanic episodes (2706 Ma) cutting older volcanic units (2734–2724 Ma), and (3) 2710 to 2704 Ma mafic to ultramafic sills intruding the 2719 to 2711 Ma episode. Changes to the nomenclature include the identification of pre-2750 Ma volcanic episode (supracrustal fragments) in the northern and southern Abitibi greenstone belt and subdivision of the 2719 to 2711 Ma, 2710 to 2704 Ma, and 2704 to 2695 Ma episodes into lower and upper parts. We present the results of this lithostratigraphic subdivision as the first geochronologically constrained stratigraphic and/or lithotectonic map of the Abitibi greenstone belt. Many of the volcanic episodes are intercalated with and capped by a relatively thin “sedimentary interface zone” dominated by chemical sedimentary rocks. Stratigraphic and geochronological analysis of these zones indicates discontinuous deposition with localized gaps of 2 to 27 m.y. between volcanic episodes. The zones consist of up to 200 m of iron formation, chert breccia, heterolithic debris flows of volcanic provenance, sandstone and/or argillite and conglomerate. Modeling of the time required for deposition of the volcanic units based on rates of magma production in modern arc and plume environments is on the order of 103 to 104 years, whereas the time interval between basalt-rhyolite cycles is 106 years. The sedimentary interface zones are therefore interpreted as condensed sections, zones with very low rates of sedimentation in a basinal setting, or zones with negligible rates of sedimentation marked by silicification of existing rock types. The sedimentary interface zones are therefore considered submarine correlative conformities, disconformities, or unconformities separating the equivalent of group level stratigraphic and lithotectonic units. The unconformity-bounded stratigraphic model provides a new regional to deposit-scale interpretive model for use in exploration for syngenetic mineralization.

All komatiitic rocks in the 2.7 Ga Abitibi greenstone belt, Canada contain similar PGE abundances, regardless of age, depth of melting, and/or residue composition. Furthermore, all komatiitic rocks in the Abitibi greenstone belt appear to have been initially sulfide undersaturated and show only local evidence of sulfide segregation, consistent with sulfide saturation during lava emplacement and/or crystallization. Ni-Cu-(PGE) mineralization in the Abitibi greenstone belt has been found only in the two oldest assemblages, the Kidd-Munro and Tisdale assemblages. Only these assemblages contain lava channels or channelized sheet flows and sulfide-bearing sediments. Thus, physical volcanology is more important than magma composition in determining the prospectivity of komatiitic rocks in the Abitibi greenstone belt and elsewhere.

Komatiitic rocks occur in four lithotectonic assemblages in the Abitibi greenstone belt of the Archean Superior province: the 2750 to 2735 Ma Pacaud assemblage, the 2723 to 2720 Ma Stoughton-Roquemaure assemblage, the 2719 to 2710 Ma Kidd-Munro assemblage, and the 2710 to 2703 Ma Tisdale assemblage. There are significant volcanological and geochemical differences between the komatiitic rocks in these assemblages. Those in the older (Pacaud and Stoughton-Roquemaure) assemblages are primarily Ti- and Al-depleted-Ti-enriched komatiites, respectively, that appear to be uncontaminated by upper continental crustal rocks, form unchannelized lava flow facies, and are not underlain by sulfur-rich metasedimentary rocks. In contrast, those in the younger (Kidd-Munro and Tisdale) assemblages are primarily Al-undepleted komatiites, locally contaminated by upper continental crustal rocks, locally form lava channels and channelized sheet flow facies, and locally overlie sulfur-rich metasedimentary rocks. Sixty-four unmineralized samples from all four komatiite-bearing assemblages in the Abitibi greenstone belt have similar unfractionated platinum-group element (PGE) abundances at 0.5 to 5 times chondritic values. Minor variations in PGE abundances are attributable to variable (but uniformly high) degrees of partial melting, fractional crystallization and/or accumulation of olivine, chromite, or clinopyroxene and, rarely, loss of sulfide or Ir-bearing alloys. Almost all of the komatiitic magmas and lavas represented by the samples, regardless of komatiite type (including Al-undepleted, Al-depleted–Ti-enriched, and Ti-depleted komatiites) and depth of melt extraction, appear to have been derived by high-degree partial melting of sources with essentially identical chalcophile element contents and to have remained undersaturated in sulfide during melt extraction, ascent, eruption, and emplacement. This is similar to most komatiites worldwide. Type I (Kambalda-type) stratiform, basal massive-disseminated and type II (Mt. Keith-type), strata-bound, internally disseminated Ni-Cu-(PGE) deposits in the Abitibi greenstone belt are hosted mainly by Al-undepleted cumulate komatiites that are associated with sulfur-rich country rocks in the Kidd-Munro (e.g., Alexo-Dundonald, Marbridge, Dumont) and Tisdale (e.g., Langmuir-Redstone, Texmont, Sothman, Bannockburn) assemblages. Although many other deposits of this type are also hosted by Al-undepleted komatiites (e.g., Damba-Silwane-Shangani, Kambalda, Mt. Keith, Perseverance), some are not (e.g., Boa Vista, Forrestania, Ruth Well), confirming that the depth of melt extraction or the precise degree of partial melting do not appear to be important controls on the genesis of magmatic Ni-Cu-(PGE) deposits. However, virtually all deposits of this type are hosted by thick, cumulate komatiite units interpreted to represent lava channels or channelized sheet flows and most are associated with sulfur-rich country rocks, suggesting that high magma and/or lava flux, lava channelization, and sulfur-rich country rocks are the most critical features in determining the prospectivity of komatiites in the Abitibi greenstone belt and elsewhere.

170 R.A. Sproule et al./Precambrian Research 126 (2003) 169–172Table 2Representative analyses of komatiites and komatiitic basalts from the Abitibi greenstone belt. Major elements are recalculated to 100% on ananhydrous basisTag# 27982 27974 23271 Roq95-10 Roq95-7 23252 23341 22179 23214Source * * * ** ** * * * *Area ShiningTreeShiningTreeShiningTreeLakeAbitibiLakeAbitibiKidd-MunroShiningTreeShawDomeRoundLakeDomeStrat. Unit Pacaud Pacaud Pacaud SR SR KM KM Tisdale TisdaleUTM N 5270125 5270350 5270360 5374000 5374000 5281266 5377720 5360755 5324512UTM E 481584.5 481450 481480 618000 618000 493967 567920 496500 568075Rock Type KomatiiticbasaltKomatiiticbasaltKomatiiteperidotiteKomatiiticbasaltKomatiiticbasaltPeridotitickomatiitePeridotitickomatiiteKomatiiticbasaltKomatiiticbasaltTexture RandomspinifexRandomspinifexOrthocumulate Aphanitic Aphanitic Orthocumulate Orthocumulate RandomspinifexRandomspinifexSiO

Komatiites and komatiitic basalts in the Abitibi greenstone belt exhibit pronounced spatial and temporal variations in major and trace element geochemistry that provide important constraints on their petrogenesis and tectonic setting. Ti-depleted komatiites exhibit high Al2O3/TiO2 (25–35) and low [Gd/Yb]MN (∼0.6–0.8), suggesting derivation by (1) dynamic melting of a refractory source leaving a refractory harzburgite residue, similar to that which produced Gorgona picrites, or (2) relatively high degree melting of a garnet-rich source leaving a peridotite residue. Al-depleted–Ti-enriched komatiites exhibit low Al2O3/TiO2 (6–14) and high [Gd/Yb]MN (∼1.2–2.0), suggesting derivation by partial melting of a garnet peridotite source leaving a garnet-bearing residue. Al-undepleted komatiites exhibit moderate Al2O3/TiO2 (15–25) and moderate [Gd/Yb]MN (∼0.8–1.2), suggesting derivation by shallower or higher-degree partial melting of a garnet peridotite source leaving a garnet-free residue. Those in the 2750–2735 Ma Pacaud assemblage are all Ti-depleted komatiites, those in the 2725–2720 Ma Stoughton–Roquemaure assemblage are predominantly Al-depleted–Ti-enriched komatiites with lesser Al-undepleted komatiites, those in the 2718–2710 Ma Kidd–Munro assemblage are predominantly Al-undepleted komatiites with lesser Al-depleted–Ti-enriched komatiites, and those in the 2710–2703 Ma Tisdale assemblage are predominantly Al-undepleted komatiites. Some of the komatiites in the younger assemblages (Kidd–Munro and Tisdale) are enriched in highly incompatible lithophile elements, exhibit low [Nb/Th]MN, and felsic-intermediate rocks in these assemblages contain inherited zircons, suggesting contamination by upper crustal rocks or sediments derived from upper crustal rocks. The high Mg contents of the komatiites suggest that they formed in mantle-derived plumes. The number of plumes required to generate the komatiitic magmatism in the Abitibi greenstone belt is not clear; it may be possible to derive all of the different magma types from different parts of a single plume at different times, but the long duration of komatiitic magmatism (≥47 My) suggests that multiple plumes are more likely. In any case, the temporal changes in composition are consistent with a decreasing influence of garnet in the source region of the plume(s) with time. There are several possible mechanisms to account of these variations, none of which are mutually exclusive: (1) a decrease in the depth of melt extraction as the plume(s) ascended; (2) a decrease in the amount of magma derived from deeper levels in the plume relative to that derived from shallower levels in the plume, possibly representing a change in plume morphology with time, for example as it was dragged by the overlying plate; (3) a decrease in the rate of plume ascent due to interaction with subducting slabs; or (4) a decrease in the thickness of the lithosphere during plume ascent, perhaps related to lithospheric thinning by the rising plume(s). The restriction of contamination to individual flow units or individual parts of flow units in the Kidd–Munro and Tisdale assemblages suggests that crustal contamination occurred locally, via assimilation of country rocks during emplacement.

Models for Superior Province greenstone belt development have changed over the last 30 years from ensimatic, generally autochthonous models to suspect allochthonous terrane models based on the plate tectonic paradigm. Models intermediate between these end-members such as autochthonous terrane models have not been prominent. Within the Superior Province, there is a secular variation from early ca. 3 Ga platformal assemblages with overlying rift-related komatiite–tholeiite units representing breakup of ca. 3 Ga continental fragments to late (ca. 2.7 Ga) pull-apart basins. Autochthonous terranes should demonstrate some sort of logical order such as that described above, whereas allochthonous terranes will be largely oceanic in character and should be essentially random in their occurrence rather than exhibiting secular order. In the face of this situation, this paper reviews evidence from throughout the Superior Province that suggests many, but not all greenstone assemblages developed autochthonously or parautochthonously. The earliest greenstone assemblages in the Superior Province are generally quartzite and carbonate-bearing platform sequences in the North Caribou terrane and the Marmion terrane deposited unconformably upon granitoid or greenstone basement. The evidence for autochthonous/parautochthonous development includes the presence of unconformities between Keewatin type greenstone units such as: (1) basal unconformities between ca. 3 Ga granitoid or volcanic basement and slightly younger platformal assemblages; (2) between ca. 2.9 Ga platformal assemblages and younger platformal and Andean arc assemblages. Stratigraphic patterns such as adjacent greenstone belt assemblages representing sharply differing geodynamic settings with some indication of palinspastic order or progression in the northern, central and southern Superior Province are considered suggestive of orderly, autochthonous progression from platforms through rifting of continental fragments and late assembly during the Kenoran orogeny. Petrogenetic data indicating involvement of older granitoid crust in genesis of greenstone assemblages of a variety of ages throughout the Superior Province suggests the presence of older basement, implying autochthonous or parautochthonous development of many assemblages. Younger greenstone assemblages in the Abitibi–Wawa subprovince and the North Caribou terrane contain xenocrystic zircons similar in age to stratigraphically underlying greenstone assemblages. This pattern has been interpreted to represent autochthonous greenstone development. The current above enumerated evidence for autochthonous development of greenstone assemblages is generally indirect; few actual unconformities between Keewatin type greenstone assemblages have been recognized in the Superior Province. To aid in future recognition of unconformities in Keewatin type greenstone assemblages, examples of unconformities of varying degrees of subtlety in Archean sequences are reviewed. Archean unconformities ranging from sharp angular discordances in the northern Superior Province to subtle stratigraphic patterns, to weathering-related geochemical mobilization along inter-formational contacts are described. These styles of unconformity in Archean sequences suggest previously unrecognized major disconformities and unconformities may well exist between Keewatin type assemblages in the Superior Province generally and the Abitibi–Wawa subprovince specifically. The most important disconformity suggested in the Abitibi–Wawa subprovince is between the ca. 2726 Ma Deloro and the ca. 2708 Ma Tisdale assemblages. The presence of unconformities and disconformities coupled with the indirect lines of evidence cited above suggest an ‘autochthonous’ origin for many Superior Province greenstone belts. This ‘autochthonous’ origin involves either autochthonous or parautochthonous evolution of assemblages at least in part in place. That is, development of ca. 3 Ga platform sequences in selected locales followed by in place evolution of a succession of plume, oceanic and arc assemblages culminating in late pull-apart basin sequences. Implications of the autochthonous model for greenstone belts of the Superior Province, particularly the Abitibi–Wawa subprovince, include the future ability to recognize unconformity bounded sequences, potentially improving regional scale correlation.

The central Wabigoon Subprovince of the western Superior Province contains Mesoarchean granitoid and supracrustal rocks (3.01–2.83 Ga) and 2.78–2.69 Ga granitoid plutons and supracrustal sequences. It is a key area for understanding the relationship between greenstone belts and surrounding granitoid rocks that may have acted as basement. The Obonga Lake greenstone belt contains two distinct assemblages: (1) a Ta>Nb>Th. They have eNd values of +0.7 to +2.4. A dacitic unit in the northern assemblage has high La/Yb, high Sr/Y, low Nb, Y and heavy REE and is interpreted to represent a mantle-modified slab melt similar to adakites in Cenozoic arcs. The associated enriched basalts may represent melts from the mantle wedge modified by slab melt (adakitic) metasomatism, or they may represent an enriched (OIB-like) asthenospheric source. Rocks of the southern assemblage mostly show Th and light REE enrichment and negative Nb and Ta anomalies. Low eNd values (down to −0.9) can be modelled through contamination of a mafic liquid by 3.3–3.2 Ga sialic crust, although rocks of that age have not been observed. Suggested Mesoarchean basement to the southern assemblage may therefore have had a complex history spanning several hundred million years. The southern assemblage volcanism may have occurred in a continental arc to continental back-arc system where crustal recycling played an important role.

The greenstone belts of the western Superior Province are predominantly 2.78 to 2.69 Ga and provide evidence of oceanic and arc volcanism during the accretionary phase of development of the Superior Province. There is also scattered evidence of Meso-Archean crust (predominantly 2.9 to 3.0 Ga) within the western Superior Province. The Meso-Archean greenstone belts commonly contain platformal sediments and unconformably overlie granitoid basement. The platformal sediments occur associated with komatiitic and tholeiitic volcanic rocks that suggest a history of magmatism associated with rifting during the Meso-Archean. The central Wabigoon Subprovince is a key area of Meso-Archean crust and in its southern portion comprises the Steep Rock, Finlayson and Lumby Lake greenstone belts. The Steep Rock greenstone belt unconformably overlies 3 Ga continental basement and contains platformal sediments succeeded by komatiitic and tholeiitic volcanic rocks. The Lumby Lake greenstone belt contains thick sequences of mafic volcanics, a number of komatiite horizons, and thin platformal sedimentary units. The two belts are joined by the predominantly mafic volcanic Finlayson greenstone belt. The volcanics throughout these three greenstone belts may be correlated to some extent and a range of basaltic and komatiite types is present. Al-undepleted komatiites present in the Lumby Lake greenstone belt have an Al2O3/TiO2 ratio ranging from 14 to 27 and (Gd/Yb)N from 0.7 to 1.3. These are divided into basaltic komatiites with generally unfractionated mantle-normalised multi-element profiles, and spinifex-textured high-Mg basalts with slightly light REE enriched multi-element profiles and small negative Nb and Ta anomalies. The unfractionated basaltic komatiites represent high degree partial melts of the upper mantle whereas the spinifex-textured high-Mg basalts represent evolutionary products of the komatiite liquids following olivine and chromite fractionation and crustal contamination. Al-depleted komatiites are present in both the Lumby Lake and Steep Rock belts and have Al2O3/TiO2 ratio ranges from 2.5 to 5. These display strong enrichment in the light REE and Nb and strong depletion in the heavy REE and Y ((Gd/Yb)N=2–4). They represent a deep mantle plume source generated from a high degree of partial melting in the majorite garnet stability field. The basaltic flows in all three greenstone belts are predominantly slightly light REE depleted and represent a slightly depleted upper mantle source. Basalts spatially associated with the unfractionated basaltic komatiites and the slightly light REE enriched spinifex-textured high-Mg basalts are also slightly enriched in light REE and have negative Nb and Ta anomalies. These basalts represent evolved products of the primitive basaltic komatiites and enriched spinifex-textured high-Mg basalts after further crustal contamination and olivine and clinopyroxene fractionation. The geochemical stratigraphy in the Lumby Lake belt is consistent with an ascending mantle plume model. The light REE depleted basalts were derived from upper mantle melted by an ascending mantle plume. These are overlain by the unfractionated basaltic komatiites and their evolutionary products which represent hotter plume head material derived from a mixture of plume mantle and entrained depleted upper mantle. In turn, these are overlain by strongly light REE and HFSE enriched komatiites that represent a deep plume source that has not been mixed with depleted mantle and are, therefore, likely to have been derived from a plume core or tail. Volcanism was protracted in these three greenstone belts lasting ca. 70 Ma and combined stratigraphic evidence from the Lumby Lake and Steep Rock belts suggests that more than one plume may have ascended and tapped the same mantle sources, over time, within the area. Plume magmatism and rifting of continental platforms thus appears to have been an important feature of crustal development in the Meso-Archean.

The Red Lake greenstone belt of the Superior Province contains a number of volcano-sedimentary assemblages that range in age from ∼ 2.99 Ga to ∼ 2.74 Ga. The oldest of these is the 2.99-2.96 Ga Balmer assemblage which is part of the North Caribou terrane, one of the oldest portions of the Superior Province. The Balmer assemblage dominates the Red Lake greenstone belt and is significant in that it contains evidence for plume-generated magmatism. The assemblage is dominated by mafic volcanic flows with intercalated komatiitic lavas, mafic and ultramafic sills and shallow water clastic and chemical sediments. The basalts have 5–12% MgO and mantle normalized signatures which range from slightly light rare earth element (REE)-depleted, through essentially flat, to light REE-enriched. Some lavas display slight negative Nb and Ti anomalies, indicating selective crustal contamination. The komatiites have 12–14% MgO, are the Al-undeplated type and are strongly depleted in light-REE. They have e Nd values of ca −2. The geochemistry, combined with SmNd isotope data, indicates that the komatiites came from a long term enriched source that was probably generated at ca 3.4 Ga. That source underwent a melting event that depleted it in light REE prior to komatiite generation. The presence of komatiites from a long term enriched source and the strong contrast in the REE profiles of the basalts and the komatiites are consistent with a plume model for the generation of the Balmer assemblage, the komatiites having originated in the plume core and the basalts from entrained upper mantle or mantle melted by the plume.

The Beardmore–Geraldton greenstone belt lies between the Wabigoon volcanic arc (Onaman–Tashota terrane) and the Quetico metasedimentary subprovince and thus has an important bearing on the accretionary model that has been proposed for the amalgamation of these terranes. This paper presents geochemical evidence for the petrogenetic affinities of the volcanic units of the western half of the Beardmore–Geraldton greenstone belt. These data suggest that the metavolcanic rocks of the greenstone belt form a series of distinct packages. Trace element data are used to demonstrate the similarities and differences of each unit of lavas and to characterize their source region and likely tectono-magmatic setting. The data indicate that three separate fragments of volcanic crust representing oceanic crust, arc crust, and back-arc crust formed in a small arc system and were juxtaposed prior to collision with the Wabigoon arc. These fragments of crust were then accreted to the Wabigoon arc where sedimentation was followed by thrusting and folding resulting in shortening of the belt. Delamination of the volcanic units is thought to have been responsible for the preservation of just the pillow lava sequence of oceanic crust. Lower crustal or crust–mantle delamination of the Wawa arc and underplating of the Quetico are thought to have been responsible for the late and long-lived, high-grade metamorphic event in the Quetico and such "flake tectonics" are thought to have been an important process in the interaction between the Wabigoon, Quetico, and Wawa subprovinces.