166-10 Improved Identification of Kimberlitic Zircons Using Trace-Element Compositions and Machine Learning: Implications for the Search for Superdeep Diamonds

Session: Mineralogical Characterization of Economic Resources: From Critical Minerals to Gemstones

Presenting Author:

Matthew Hardman

Authors:

Hardman, Matthew F.¹, Pearson, D. Graham², DuFrane, S. Andy³, Cabral-Neto, Izaac⁴, Azzone, Rogério Guitarrari⁵, Shu, Qiao⁶, Hinde, Jason⁷, Rukhlov, Alexei S.⁸

(1) Gemological Institute of America, Carlsbad, California, USA, (2) University of Alberta, Earth & Atmospheric Sciences, Edmonton, Alberta, Canada, (3) University of Alberta, Earth & Atmospheric Sciences, Edmonton, Alberta, Canada, (4) Geological Survey of Brazil - SGB/CPRM, Natal, Rio Grande do Norte, Brazil; Institute of Geosciences, University of São Paulo, São Paulo, São Paulo, Brazil, (5) Institute of Geosciences, University of São Paulo, São Paulo, São Paulo, Brazil, (6) University of Alberta, Earth & Atmospheric Sciences, Edmonton, Alberta, Canada; State Key Laboratory of Ore Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, Guizhou, China, (7) University of Alberta, Earth & Atmospheric Sciences, Edmonton, Alberta, Canada; APEX Geoscience Ltd., Edmonton, Alberta, Canada, (8) BC Geological Survey, Victoria, BC, Canada,

Abstract:

Zircon is an important pathfinder mineral for identifying igneous source rocks, including kimberlite and carbonatite. However, discrimination of zircons from these two lithologies is challenging due to their surprisingly limited published trace element data and their compositional overlap. Kimberlites can be diamond-bearing, and in rare cases these diamonds may be sourced from the sublithospheric mantle. Such diamonds are often sought after due to their size and purity. Consequently, enhanced tools to identify kimberlite deposits containing superdeep diamonds can provide new value to diamond exploration practices.

Here we determined the trace-element compositions of 170 new zircon megacrysts from kimberlites and of 220 new zircons from global carbonatites and related rocks. The kimberlitic zircon megacrysts in the present study have a relatively narrow range of trace-element compositions whereas the new carbonatite zircons are compositionally diverse and likely reflect formation under varied geological conditions from a variety of heterogeneous sources, as well as complex equilibrium mineral assemblages. We have also compiled zircon trace-element compositions for a wide variety of rocks from published literature (including from crustal rocks), allowing for more general classification approaches.

We apply random forest and discriminant projection analysis to distinguish zircons from kimberlite and carbonatite from those in many crustal lithologies. Both approaches can discriminate kimberlite-related zircons from those in other host rocks, with error rates as low as 8.5±10.0% (2σ). We further apply our database to evaluate whether there is a compositional fingerprint that can distinguish zircon megacrysts from kimberlites containing superdeep diamonds from those that do not. While we find broad separation between subsets of zircon from kimberlites with and without superdeep diamonds using our preliminary dataset, more compositional data for zircons from known superdeep-diamond-bearing kimberlites are necessary to evaluate any possible compositional fingerprint. These new data do, however, allow broader investigation of the petrogenesis of zircon megacrysts in kimberlites.

Geological Society of America Abstracts with Program. Vol. 57, No. 6, 2025

doi: 10.1130/abs/2025AM-9282

© Copyright 2025 The Geological Society of America (GSA), all rights reserved.