TY - JOUR
T1 - A new approach to reconstructing the composition and evolution of kimberlite melts
T2 - a case study of the archetypal Bultfontein kimberlite (Kimberley, South Africa)
AU - Soltys, Ashton
AU - Giuliani, Andrea
AU - Phillips, David
PY - 2018/4
Y1 - 2018/4
N2 - The compositions of kimberlite melts at depth and upon emplacement in the upper crust remain elusive. This can be attributed to the unquantified effects of multiple processes, such as alteration, assimilation, xenocryst contamination, and fractional crystallisation. The inability to accurately constrain the composition and physical properties of kimberlite melts prevents a comprehensive understanding of their petrogenesis. To improve constraints on the compositions of kimberlite melts, we have combined modal analysis including the discrimination of xenocrystic from magmatic phases, with mineral chemistry determinations to reconstruct a whole-rock composition. We apply this approach to a sample of "fresh" macrocrystic hypabyssal kimberlite (sample BK-1) from the Bultfontein mine (Kimberley, South Africa). The accuracy of this whole-rock reconstruction method is validated by the similarity between reconstructed and measured whole-rock compositions. A series of corrections are then applied to account for the effects of post-emplacement serpentinisation, pre-emplacement olivine crystallisation, and the inclusion and assimilation of mantle material. This approach permits discernment of melt compositions at different stages of kimberlite evolution. The primitive melt parental to the Bultfontein kimberlite is estimated to contain 17.4–19.0 wt% SiO2, 20.2–22.8 wt% MgO, 20.9–21.9 wt% CaO, 2.1–2.3 wt% P2O5, 1.2–1.4 wt% TiO2, 0.9–1.1 wt% Al2O3, and 0.6–0.7 wt% K2O, and has a Mg# of 83.4–84.4. Primary volatile contents (i.e., after an attempt to account for volatile loss) are tentatively estimated at ~2.1–2.2 wt% H2O and ~22.9–25.4 wt% CO2. This composition is deficient in SiO2, MgO and H2O, but enriched in CaO and CO2 compared with most previous estimates of primitive kimberlite melts. We suggest that the primitive melt parental to the Bultfontein kimberlite was a transitional silicate-carbonate melt, which was progressively enriched in SiO2, MgO, Al2O3, Cr2O3, and Na2O through the assimilation of lithospheric mantle material. Comparisons with experimentally produced low-degree melts of carbonated lherzolite indicate that the Bultfontein kimberlite could have formed by ~0.5% melting of asthenospheric mantle at ~6.0–8.6 GPa (i.e., ~190–285 km) and ~1400–1500 °C. The low calculated Na2O contents (<0.2 wt%), which are inconsistent with derivation from low-degree melting of lherzolite, suggest that an alkali-bearing, volatile-rich fluid was exsolved during ascent or released after emplacement, and subsequently removed.
AB - The compositions of kimberlite melts at depth and upon emplacement in the upper crust remain elusive. This can be attributed to the unquantified effects of multiple processes, such as alteration, assimilation, xenocryst contamination, and fractional crystallisation. The inability to accurately constrain the composition and physical properties of kimberlite melts prevents a comprehensive understanding of their petrogenesis. To improve constraints on the compositions of kimberlite melts, we have combined modal analysis including the discrimination of xenocrystic from magmatic phases, with mineral chemistry determinations to reconstruct a whole-rock composition. We apply this approach to a sample of "fresh" macrocrystic hypabyssal kimberlite (sample BK-1) from the Bultfontein mine (Kimberley, South Africa). The accuracy of this whole-rock reconstruction method is validated by the similarity between reconstructed and measured whole-rock compositions. A series of corrections are then applied to account for the effects of post-emplacement serpentinisation, pre-emplacement olivine crystallisation, and the inclusion and assimilation of mantle material. This approach permits discernment of melt compositions at different stages of kimberlite evolution. The primitive melt parental to the Bultfontein kimberlite is estimated to contain 17.4–19.0 wt% SiO2, 20.2–22.8 wt% MgO, 20.9–21.9 wt% CaO, 2.1–2.3 wt% P2O5, 1.2–1.4 wt% TiO2, 0.9–1.1 wt% Al2O3, and 0.6–0.7 wt% K2O, and has a Mg# of 83.4–84.4. Primary volatile contents (i.e., after an attempt to account for volatile loss) are tentatively estimated at ~2.1–2.2 wt% H2O and ~22.9–25.4 wt% CO2. This composition is deficient in SiO2, MgO and H2O, but enriched in CaO and CO2 compared with most previous estimates of primitive kimberlite melts. We suggest that the primitive melt parental to the Bultfontein kimberlite was a transitional silicate-carbonate melt, which was progressively enriched in SiO2, MgO, Al2O3, Cr2O3, and Na2O through the assimilation of lithospheric mantle material. Comparisons with experimentally produced low-degree melts of carbonated lherzolite indicate that the Bultfontein kimberlite could have formed by ~0.5% melting of asthenospheric mantle at ~6.0–8.6 GPa (i.e., ~190–285 km) and ~1400–1500 °C. The low calculated Na2O contents (<0.2 wt%), which are inconsistent with derivation from low-degree melting of lherzolite, suggest that an alkali-bearing, volatile-rich fluid was exsolved during ascent or released after emplacement, and subsequently removed.
KW - Kimberlite
KW - Melt evolution
KW - Primitive Kimberlite melt
KW - Melt reconstruction
UR - http://www.scopus.com/inward/record.url?scp=85044359446&partnerID=8YFLogxK
UR - http://purl.org/au-research/grants/arc/DE150100009
U2 - 10.1016/j.lithos.2018.01.027
DO - 10.1016/j.lithos.2018.01.027
M3 - Article
AN - SCOPUS:85044359446
SN - 0024-4937
VL - 304-307
SP - 1
EP - 15
JO - Lithos
JF - Lithos
ER -