Metallogenesis of the Cu-Co deposits in the Central African Copperbelt, information from REE patterns and the Lu-Hf isotopic system.

This PhD thesis is centered on the Central African Copperbelt, a Neoproterozoic stratiform Cu-Co ore province, and consists of three main parts. The first part focuses on the geotectonic evolution of the basement units in the vicinity of the Katanga basin that hosts the ore deposits. The second part uses rare earth element patterns in gangue carbonates associated with the ores as proxies for hydrothermal mineralization-related processes. The final part assesses the dating and metal source tracking potential of the Lu-Hf isotope system for hydrothermal carbonates.

Part I. Geotectonic history of the Central African cratons that surround the Neoproterozoic Copperbelt ore province

The Central African Copperbelt is one of the three supergiant sediment-hosted copper provinces, along with the Siberian Kodaro-Udokan and the Central European Kupferschiefer basins. The favorable architecture and metal availability that resulted in the formation of these giant ore provinces is ultimately governed by the tectonic evolution that preceded the mineralization. Therefore, the first stage of this project examined the tectonic history of the basement units surrounding the Neoproterozoic Copperbelt, using whole rock geochemical data from representative igneous and metasedimentary samples taken from basement units surrounding the Copperbelt, and combining this data with Sr-Nd isotopic results.

In the Domes inliers directly below the Copperbelt, depleted mantle Nd model ages indicate crustal residence times of 2.4-2.6 Ga for granitoids in the eastern Domes region, while ortho- and paragneisses in the central Mwombezhi Dome have variable model ages of 2.2-3.1 Ga. The elevated model ages preclude a Paleoproterozoic juvenile island arc setting and show that these inliers contain cryptic Archean sources, both in the central Mwombezhi Dome and in the eastern part of the Domes region. These crustal residence times are broadly similar to those in the adjacent Mansa region in the Bangweulu Belt where metavolcanites give model ages of 2.2-2.3 Ga and granitoids have ages around 2.5 Ga. In the Central Bangweulu region, granitoids have crustal residence times of 2.9, 3.1 and 3.6 Ga, while metasedimentary rocks show crustal residence times of 1.9-2.6 Ga. The Irumide Belt shows crustal residence times between 2.5 and 3.2 Ga for both igneous and metasedimentary rocks, reinforcing its proposed origin as a metacratonized margin of the Bangweulu Block. Our data for the Bangweulu and Irumide regions thus confirms the abundant reworking of crustal material since the Archean. The petrochemical and isotopic similarities between the 1880-1850 Ma igneous rocks in the Mansa region and the granitoids of the eastern Domes region corroborate the proposed north-south oriented convergent setting for the Paleoproterozoic magmatism in this region.

The geochemical composition of the mafic Kibara Belt rocks is indicative for an arc-type setting, which in combination with  , elevated crustal residence times of 2.7-2.8 Ga and a minor suite of weakly metaluminous granites confirms the previously inferred Mesoproterozoic active margin setting. An active margin setting in the Kibara Belt is difficult to reconcile with the contemporaneous intracratonic setting envisaged for the adjacent Karagwe-Ankole Belt, since both Mesoproterozoic belts are bound by the Congo and Tanzania-Bangweulu plates. Based on the currently available geochemical, radiometric and structural data, we proposed a regional geodynamic framework that includes a transition from an extensional to a collisional regime in the Karagwe-Ankole Belt. This framework accounts for the presence of arc-type meta-volcanites in the Karagwe-Ankole Belt and a sedimentary hiatus at 1375-1220 Ma. This implies a back arc basin nature for the Karagwe-Ankole Belt, superimposed on a large igneous province evidenced by the c. 1370 Ma Lake Victoria Dyke Swarm. The 1400-1375 Ma ultramafic layered complexes along the Kabanga-Musongati alignment are likely related to asthenospheric upwelling along a transcrustal weakness zone, which is currently interpreted as a paleo-suture.

The composition of the strongly peraluminous granites in the Kibara Belt is not indicative for a specific tectonic setting, but is consistent with metapelite dehydration melting. Similar isotope characteristics as the older granites suggest that the 1375 Ma and 1000 Ma suites melt similar source material. The precise tectonic setting of the 1000 Ma magmatic suite in the Kibara Belt remains enigmatic. Mantle delamination would generate the required heat for voluminous granite intrusions, but would be associated with minor mafic magmatism. The current evidence favors a collisional setting, which would also account for the occurrence of orogenic gold deposits.

Part II. Systematics of rare earth element and yttrium (REEY) patterns in gangue carbonates: potential for tracking metal sources and hydrothermal processes

The metals in the Copperbelt ores are thought to be mainly leached from the basement rocks by hot brines moving through the subsurface along fractured zones. The resulting metal-rich brines subsequently interacted with sedimentary units rich in sulfides, thereby generating metal-sulfide deposits. The second part of the project focuses on the potential of REEY patterns in gangue carbonates associated with the ore sulfides for tracing metal sources and hydrothermal processes. In a first stage, source-sink approach is adopted to visualize the main processes that control hydrothermal mass transfer of rare earth elements and yttrium. The REEY sourcing during fluid-rock interaction is a complex function of mineral stability at the interaction conditions, and the currently available approximations for this process are mostly empirical. Along the fluid pathway, the REEY fractionation is also controlled by complexation- and sorption-driven fractionation, which further influences the saturation state of REEY-sequestering mineral phases. Rayleigh fractionation of these mineral phases along the pathway may fractionate the REEY, depending on the mineralogical control at the precipitation conditions. At the deposit site, the REEY are redistributed according to the mineralogical control of the gangue carbonate and the coprecipitating phases, and the retention in stable complexes in the fluid. This systematic approach reveals that the REEY patterns of hydrothermal carbonates are a complex function of superimposed processes for which quantitative constraints are mostly lacking. While the resulting potential for source tracking is limited, REEY patterns in gangue carbonates can be used to identify the dominant hydrothermal processes along the fluid pathway that shaped the patterns at the deposit site.

This approach has been applied to six case study deposits, to test the deposit-scale variation in REEY fractionation by hydrothermal processes at different stages of the multi-phase mineralization, and to verify the presence of a lithology-dependent host rock effect.

The Kamoto and Luiswishi deposits are two Co-rich Congo-type stratiform deposits hosted mainly by carbonate rocks. The gangue dolomites associated with the ores were characterized by an extreme depletion in light rare earth elements (LREE), when normalized over the Post-Archean Average Shale Composition (PAAS), with upwards convex pattern shapes and relatively low total rare earth element concentrations. The upwards convex pattern shape is more pronounced in the Kamoto dolomites compared to Luiswishi dolomites and the latter tend to show more pronounced negative Eu anomalies. Magnesites at Luiswishi show distinct, log-linearly increasing patterns, consistent with the inferred heavy rare earth element (HREE) selectivity of magnesite. The differences between the dolomite patterns at both deposits in terms of Eu anomalies and mid to heavy REE fractionation can be attributed to different physicochemical conditions during mineralization, as indicated by the higher metamorphic grade at Luiswishi. In general, the pattern characteristics at these deposits are indicative for precipitation of an LREE selective mineral phase along the fluid pathway, possibly a phosphate or a carbonate mineral.

The Nkana and Konkola deposits are Zambia-type stratiform deposits with high and low Co content respectively, and are hosted mainly by siliciclastic rocks, with variable proportions of carbonate. The REEY patterns from Nkana gangue dolomites and calcites evolve from upwards-convex PAAS-normalized to log-linearly increasing pattern shapes. Together with the petrological and structural characteristics, this change in pattern shape is interpreted to reflect an evolution from relatively low temperatures to greenschist facies conditions at the deposit site. The LREE depletion at Nkana and Konkola is not as pronounced compared to the Kamoto and Luiswishi dolomites and the total REE content is generally higher. The differences between the studied Zambia-type and Congo-type deposits are therefore suggestive of a lithology-dependent host rock effect.

The upwards convex PAAS-normalized patterns in gangue carbonates from the vein-type Dikulushi base metal deposit are consistent with a remobilization of the early Cu-Pb-Zn mineralization phase by a later Cu-Ag phase, as indicated by the similar pattern shapes in calcites associated with both phases. The lower REEY concentrations in calcites associated with the Cu-Ag phase indicate conditions that are less favorable for REEY transport. The vein-type Kipushi gangue dolomites show either flat to upwards-convex patterns without Eu anomalies and with low REEY concentrations or strongly LREE-depleted upwards-convex patterns with pronounced negative Eu anomalies and higher overall REEY concentrations. Combined with the major element composition in these respective pattern types, this is interpreted to record a transition from reducing towards more oxidizing conditions in the mineralization fluid.

Part III. Dating potential of the Lu-Hf and Sm-Nd methods for hydrothermal carbonates

The final part of this project focused on the dating potential of the Lu-Hf and Sm-Nd system for hydrothermal carbonates. Both dating methods were tested on gangue dolomites from the Kipushi deposit. Hf could not be reliably detected within the spiked fractions or in the unspiked fractions. In practical terms, the low Hf concentrations imply that preconcentration methods are likely a prerequisite for successful Lu-Hf dating of hydrothermal carbonates. The Sm-Nd results suggest that the samples contained similar initial Nd isotope compositions within error, but more data with larger spread in Sm/Nd ratios is needed to confirm the dating potential of this method.