Controls on the explosive emplacement of diamondiferous kimberlites: New insights from hypabyssal and pyroclastic units in the Diavik mine, Canada

Abstract

Kimberlites are mantle-derived magmas that either crystallise as hypabyssal intrusions, erupt explosively after rapid ascent to the surface, or less commonly form lava lakes and flows, thereby creating texturally distinct kimberlite units. Efforts to fully understand the processes responsible for the explosive eruption of kimberlite magmas have been hindered by the widespread alteration and crustal contamination of most volcaniclastic kimberlites. To address this issue, we have undertaken a detailed petrographic and mineral chemical study of fresh (i.e. minimally altered) pyroclastic and hypabyssal kimberlites (HK) from the ca. 55-56 Ma A154 North and South kimberlite pipes in the Diavik Mine (Lac de Gras, Canada). These localities host exceptionally fresh kimberlites and are therefore ideally suited to this study.Kimberlite emplacement at A154 North and South initiated with the intrusion of hypabyssal kimberlite (external dykes), and was followed by the explosive formation of kimberlite pipes and volcaniclastic kimberlite infill. Subsequent kimberlite magmas intruded the volcaniclastic kimberlite units forming multiple cross-cutting, internal dykes. The studied volcaniclastic units feature abundant rounded magmaclasts and massive textures, suggestive of primary deposits. These units are classified as pyroclastic kimberlites (PK).Pyroclastic and hypabyssal kimberlite units at Diavik exhibit subtle mineral compositional differences. Samples from both internal HK units and PK units feature identical compositions for liquidus olivine rims (Mg# = 90.5 ± 0.1 and 90.7 ± 0.2, respectively), with a marginally lower Mg# of 90.2 ± 0.2 in olivine rims from the external HK dykes. Similarly, early-formed chromite compositions are the same for internal HK and PK units (Cr# = 79.1 ± 3.4 and 78.3 ± 5.7; Mg# = 60.0 ± 1.3 and 60.0 ± 2.2), but, differ in the external HK units (Cr# = 86.9 ± 2.7; Mg# = 52.8 ± 1.9). The internal HK and PK units also exhibit lower carbonate contents than the internal HK units. These compositional differences indicate that the external dykes were probably derived from slightly different primitive melt compositions to those parental to the internal HK and PK units. Spinel evolutionary trends from chromite to magnesian ulvӧspinel-magnetite (MUM) compositions are indistinguishable in internal HK and PK samples (Fe3+# = 47.2 ± 5.8 and 49.7 ± 9.3; Cr# = 25.7 ± 11.0 and 17.0 ± 14.0). These results demonstrate that the primitive melt compositions and early magmatic evolution processes are identical for the internal kimberlite units, regardless of whether the kimberlite melts erupted explosively or were emplaced as shallow intrusions. However, magmaclasts in the PK units contain higher abundances of phlogopite (<52 vol. %) and lower quantities of carbonate (<4 vol. %) than the groundmass of the hypabyssal kimberlite samples (<2 vol. % and 25–65 vol. %, respectively). This indicates that the explosively erupted magmas featured higher H2O/CO2 ratios. In contrast, abundant carbonates, including dolomite, in the internal HK samples indicate that CO2, and therefore low H2O/CO2 ratios, were retained during the emplacement of this magma, which likely prevented phlogopite crystallisation. Lower K and Rb whole-rock compositions for internal HK samples compared to PK samples, are attributed to the removal of these components in late-stage kimberlitic fluids, as indicated by hydrothermal alteration of the adjacent volcaniclastic kimberlite units. The above results clearly rule out variations in primitive melt composition and melt evolution trajectories as a primary control on the explosive behaviour of the kimberlite magmas at Diavik. Controls on kimberlite explosivity at Diavik are likely due to external factors, such as local stress regimes, the availability of groundwater (i.e. phreatomagmatism) and differing magma supply rates.