A newly recognised mafic sill-hosted Ni-sulfide deposit emplaced during the 2.4 Ga Widgiemooltha dike swarm event, Eastern Goldfields, Western Australia

Abstract

The Cathedrals Ni-Cu prospect, located at the western margin of the Eastern Goldfields of the Yilgarn Craton, is hosted within a mafic intrusion interpreted as a sill complex. U-Pb dating of apatite from the sill yielded a crystallisation age of 2336 ± 64 Ma, inferring an association of sill emplacement and Ni mineralisation related to emplacement of the c. 2400 Ma Widgiemooltha dike swarm. The sill is typically differentiated into a lower olivine orthocumulate layer overlain by a dolerite unit containing xenoliths of partially assimilated granitoids in its upper portion. The latter is interpreted to be the result of stoping and melting of the granitic hanging wall, thereby creating a gravitationally stable buoyant melt layer beneath the top contact. Ni-Cu-Fe sulfides are increasingly abundant towards the base of the sill, ranging from globular disseminated sulfides to net-textured and massive sulfides at the basal contact. The presence and orientation of sulfide globule-bubble pairs indicates a primary near-horizontal orientation. Massive sulfides commonly exhibit a loop texture with pyrrhotite grains surrounded by pentlandite and chalcopyrite. Despite the variety of sulfide textures, sulfur isotopes have a homogeneous mantle-like signature without significant mass independent fractionation. Mineral chemistries that indicate sulfide prospectivity in larger intrusions do not work as effectively in this small sill, therefore new indicators may need to be developed to explore for similar deposits. To date, there are no other known magmatic deposits of this age in Australia. Sills of this age may be more prospective than previously recognised.

Introduction

Australia is a major contributor to the world’s Ni supply, ranking third in total Ni reserves (Hoatson et al. 2006; Mudd and Jowitt 2022). The majority of Australia’s Ni reserves are within Western Australia. These are largely hosted in Archean komatiites (e.g. the 2.7 Ga Kambalda, Perseverance & Mount Keith deposits; Lesher et al. 1984; Barnes 2006; Gole and Barnes 2020) with a minor proportion associated with Proterozoic mafic-ultramafic intrusions of tholeiitic character such as Nova-Bollinger (Taranovic et al. 2022), Nebo-Babel (Godel et al. 2011) and Savannah (Le Vaillant et al. 2020). Magmatic nickel sulfides are well known to be associated with large volumes of magma with world’s largest deposits hosted in mafic Large Igneous Provinces (LIP) (e.g., the Norilsk-Talnakh deposits in Russia associated with the Siberian traps; Barnes et al. 2020a). Mafic LIPs are represented by continental flood basalts, mafic sill complexes, mafic-ultramafic layered intrusions, and mafic dike swarms (Pirajno and Hoatson 2012; Ernst 2014). Giant radiating mafic dike swarms are part of the plumbing systems of LIPs and their orientations indicate the location of the plume centre (Ernst and Buchan 1997; Buchan and Ernst 2021). There are rare occurrences of mafic dike swarms hosting magmatic sulfide mineralisation such as the platinum group element (PGE)-producing Great Dyke in Zimbabwe (Wilson and Prendergast 1989) and the Muskox intrusion in the Mackenzie dike swarm in northern Canada (Chamberlain 1967), although economic Ni-Cu dominant deposits in this setting are uncommon.

In Western Australia, minor Ni-Cu sulfide occurrences are known within the E-NE trending Jimberlana Dyke and the E-W orientated Celebration Dyke, both members of the c. 2400 Ma Widgiemooltha Dolerite, part of the Widgiemooltha dike swarm that extends across much of the Yilgarn Craton (Fig. 1). The Cathedrals belt magmatic nickel sulfide prospect, currently explored by St George Mining Limited (ASX: SGQ), comprises a series of Ni-bearing orebodies containing both disseminated and massive sulfides that occur along a 5.5 km east-west strike length (Fig. 2). No previous studies of this prospect have been published. The intersection of massive sulfides within an intrusion of this type in the Eastern Goldfields is unusual, and potentially indicative of a new search area for magmatic nickel sulfides. Understanding the ore formation and emplacement mechanisms and confirming its potential affiliation with the Widgiemooltha dike swarm are therefore key for future exploration targeting in the region.

Fig. 1

Interpreted geological map of Western Australia (modified from Geological Survey of Western Australia, 2020) illustrating Ni metallogenic events with a focus on the Widgiemooltha dike swarm and Ni occurrences. A is Agnew-Wiluna Belt, U is Ularring Belt, M is Menzies Belt

Fig. 2

a. Regional geological map (modified from Geological Survey of Western Australia 2020): b. Regional reduced to pole aeromagnetic image showing the trace of the dated Warakurna sill, which cuts through the Sandstone greenstone belt. The location of the sample from the sill dated at c. 1070 Ma is shown by the green star. Other dates shown are from granite outcrops (Compilation of geochronology information, 2020). Also shown are the Widgiemooltha dikes and 1:500k-scale structures sourced from GSWA. c. Reduced to pole aeromagnetic image with interpreted structures and mafic dikes or sills focusing on the Cathedrals belt

In this study, we explore several prospects within the Cathedrals belt, E-W orientated mafic intrusion (Figs. 1 and 2). New mineralogical data and geochemical characterisation of this orebody are provided to understand ore formation and emplacement, as well as new age constraints. A range of techniques were utilised to carry out multiscale characterisation, including: (a) the specialist Maia Mapper µ-XRF mapping instrument where up to half meter core lengths can be scanned for elemental compositions at 30 μm resolution; (b) Maia detector using the X-ray fluorescence microscopy beamline (XFM) at the Australian synchrotron; (c) automated mineralogy with the Tescan’s Integrated Mineral Analyser (TIMA), SEM-EDS-CL, and optical microscopy for identifying ore-bearing minerals, mineral assemblages and magmatic and hydrothermal alteration textures; (d) high-resolution X-ray computed tomography (HRXCT) to evaluate the 3D geometry of sulfides at the transition from disseminated to net-textured and massive sulfides; (e) LA-ICP-MS for mineral chemistry and U-Pb dating, and (f) in situ sulfur isotopes measured by secondary-ion mass spectrometry (SIMS). Combined with drill core information, these multi-scale techniques provide an efficient way to collect significant volumes and variable types of data that can be screened for analysis. This is required to test whether mineral indicators can be used for exploration in the setting examined, and their relationships to Ni sulfide mineralisation. The data also provide information about the orientation and architecture of the mafic intrusion, as well as constraining the timing of emplacement.

Geological background

The study area lies to the east (within the hanging wall) of the northern portion of the Ida Fault, which separates the Kalgoorlie Terrane of the Eastern Goldfields Superterrane from the Youanmi Terrane (Zibra et al. 2020). It is located about 120 km south-southwest of the Agnew-Wiluna belt, and at the northern end of the northwest-trending Ularring greenstone belt, which is intruded and truncated by granite to the north (Figs. 1c and 2). The granite is intruded by Proterozoic dikes or sills.

Ularring Greenstone belt

The Ularring greenstone belt (Zibra et al. 2020), historically known as the Mount Ida greenstone belt (Wyche 1999, 2003; Mahon 2021), comprises a thick succession of Kalgoorlie Group metabasalt and minor metasedimentary rocks overlain by felsic metavolcanic rocks. These appear to host partially serpentinised, pyroxene-dominated ultramafic units interbedded with sulfidic chert and shale and locally, gabbroic rocks (O’Neill 2006; Mahon 2021). Greenstones in the footwall of the Ida Fault are dominated by metabasalt interlayered with Banded Iron Formation (BIF), which are more typical of greenstones of the Youanmi Terrane (Zibra et al. 2020). The ultramafic rocks have been divided into three informal belts—Western, Central and Eastern—primarily based on their strong magnetic response (Fig. 2; O’Neill 2006). The Western unit has an overall apparent lower magnetic response, although it includes abundant remanent anomalies. It also appears to extend farther northwest into the main granite intrusion than the other belts. The central ultramafic unit is described as komatiitic flows and sills with an estimated maximum thickness of about 800 m, whereas the easternmost unit is interpreted to be about 400 m thick (O’Neill 2006). Much of the succession of the central unit is overprinted by bladed tremolite-chlorite with poor preservation of primary textures (O’Neill 2006).

The greenstone succession has been intruded by granite and pegmatite veins. None of these rocks have published dates, however granite gneisses in the region are interpreted to have magmatic crystallization ages between about 2690 and 2667 Ma, whereas undeformed granites are interpreted to have magmatic crystallization ages between 2665 and 2632 Ma (Zibra et al. 2020 and references therein).

The region around the Ularring and neighbouring greenstone belts has undergone several deformation episodes. The most recognised are: D1_e, a primary extensional phase during deposition of the main greenstone sequences; D1, a contractional episode between 2685 and 2675 Ma; D2, a major contractional episode at c. 2670 to 2665 Ma; D3 and D4, several transcurrent events that were complete by 2630 Ma (Blewett et al. 2010; Zibra et al. 2020). Dextral movement along the Waroonga Shear Zone, which truncates the Ida Fault, is interpreted to have occurred at c. 2660 Ma (Zibra et al. 2017).

Cathedrals mafic intrusion

North of the Ularring greenstone belt is an east-west trending belt of undated mafic intrusions emplaced into Archean granite. These intrusions lie parallel to a series of faults or shear zones, several of which can be interpreted as having dextral kinematics denoted by foliation and magnetic feature deflection. Recent acquisition and interpretation of seismic data by St George Mining Limited shows a series of south-dipping faults that post-date the moderately north-dipping Cathedrals Fault and another north-dipping fault north of this (St George Mining Limited 2021). The Cathedrals Fault is the major structure in the area and is associated with a mafic-ultramafic sill and a series of Ni-Cu-PGE deposits and prospects. No resource estimates are available, but the company reports 11 m of nickel-copper sulfides intersected at the Investigator prospect with concentrate grades of 11.5% Ni, 3.32 g/t Pd, 0.65 g/t Pt (St George Mining Limited 2022). Calculated sulfide tenors (contents in 100% sulfides) using assay data from St George Mining Ltd indicate median tenors of 600 ppb Pt, 2,600 ppb Pd, 6 wt% Ni and 2.5 wt% Cu in the massive sulfides (Fig. 3). PGE tenors in the disseminated sulfides are higher with median values of 1,500 ppb Pt and 6,200 ppb Pd.

Fig. 3

Boxplot illustrating the sulfide tenors calculated for the Cathedrals belt ores

Several granite gossans aligned east-northeast were discovered in 1974 at what is now the Cathedrals prospect, close to a dolerite dike (Fig. 2c; Donaldson (1974) in Parola (2015)). These were described as follows: “The granite-hosted massive Fe-Ni sulfide gossans occur as 5 small rubble heaps to the east of the Western ultramafic, near a dolerite dike that cuts the ultramafic unit.” (O’Neill 2006). Rafts of ultramafic rock within the granite have also been described and were confirmed by drilling at the Cathedrals prospect by BHP Billiton Nickel West (Parola 2015). They described the ultramafic body at Cathedrals as a meso-cumulate peridotite containing disseminated, net-textured and massive nickel sulfides, potentially incorporated into an east-west shear zone during emplacement of the granite. They commissioned Mira Geoscience to model geophysical data in 3D to investigate the likelihood of continuation of the northwest-trending ultramafic belts beneath the granite. Their results showed that this was unlikely. Since acquisition of tenements in the project area, St George Mining Ltd has discovered additional prospects within the Cathedrals belt and exploration is ongoing. The exploration history of the project area is provided in Mahon (2021).

Proterozoic dike events

The age of the Cathedrals mafic intrusion was unknown prior to this study. Known mafic magmatic intrusions other than greenstones proximal to the study area include those of the 2408–2401 Ma Widgiemooltha Large Igneous Province (LIP), of which the Widgiemooltha dyke swarm is a major part, the c. 1210 Ma Marnda Moorn LIP, and the c. 1075 Ma Warakurna LIP (Wingate 2017). The Widgiemooltha dike swarm transects much of the Yilgarn Craton and is defined by intrusions of the Widgiemooltha Dolerite, including dikes that are several 100s of km long and relatively thick. These include the well-known Jimberlana Dyke that crops out near Norseman and contains both mafic and ultramafic assemblages (Campbell et al. 1970). Although voluminous in terms of the dike swarm, not much is known about the tectonic setting and impact of the Widgiemooltha LIP on the Yilgarn Craton and the magmatic event remains enigmatic.

The Widgiemooltha Dolerite dikes trend about 075° to 085°, exhibit both positive and negative magnetic polarities and are dated at 2408 ± 3 Ma (Wingate 2017) and 2401 ± 1 Ma (Pisarevsky et al. 2015), respectively (Fig. 2b). The circa 10 million year difference in age and change in polarity are interpreted to reflect a reversal of the geomagnetic field and movement of the Yilgarn Craton towards the geographic pole (Pisarevsky et al. 2015). The Widgiemooltha dike swarm (and LIP) has been interpreted as linked to plume activity and the NW-NNW-trending dyke swarm in the Zimbabwe Craton via paleomagnetic data and a date of 2408 ± 2 Ma for the Sebanga Poort dike (Söderlund et al. 2010; Pisarevsky et al. 2015). The reconstruction shown in Pisarevsky et al. (2015) shows the Great Dyke layered intrusion of the Zimbabwe Craton, part of the c. 2575 Ma Great Dyke LIP (Söderlund et al. 2010) potentially linking to the unexposed or buried southern edge of the Yilgarn Craton (now the Southern Ocean).

The Marnda Moorn LIP includes the Muggamurra Dolerite dikes of the Marnda Moorn Suite which occur northwest of the study area and typically trend east-northeast to northeast (Wingate and Pidgeon 2005; Wang et al. 2014). The Warakurna LIP includes dikes and sills of the Warakurna Supersuite that typically occur as fine- to medium-grained dolerite and gabbro with an east-west trend (Wingate et al. 2004). A large, gently north-dipping sill that has a distinctive trace on aeromagnetic images occurs northwest of the study area (Fig. 2b). An outcrop of medium- to coarse-grained gabbro from this sill yielded baddeleyite grains which have been dated at 1070 ± 18 Ma (green star on Fig. 2b; Wingate et al. 2008), indicating a Warakurna affinity. Similar sills, or occurrences of the same sill may occur to the south (Wingate et al. 2008), potentially within the Cathedrals belt.

Methodology

In this study, we sampled 57 core intervals from nine drill cores selecting 62 domains for automated mineralogy (TIMA) analyses (18 polished thin-sections and larger blocks, and 44 polished round blocks of 25 mm diameter). Six samples were also selected for bright phase search analyses to identify and locate potential phases for U-Pb dating. Two 0.3 m half-cores were selected for Maia mapping to locate As-rich minerals and test their use as a mineral indicator, and to identify and locate minerals to be dated. The composition of spinel (from 15 samples), and silicate minerals such as olivine, clinopyroxene and plagioclase (from 5 samples) were measured by LA-ICP-MS. Additionally, we acquired a laser-ablation ICPMS map of a large euhedral chromite. Olivine and chromite from one sample were also analysed with electron microprobe (EPM). U-Pb dating of zircons sourced from the granitic host-rock was undertaken on 11 samples, and seven samples from within the mineralised intrusion were selected for U-Pb dating of apatite. A 12 cm-long drill core sample illustrating the transition from disseminated sulfides, net-textured sulfides to massive sulfides was analysed using high-resolution X-ray computed tomography to evaluate the 3D textures of sulfides. The sulfur isotope composition of pyrrhotite, pentlandite, chalcopyrite and pyrite from this sample and a chilled margin were also analysed.

TIMA

Quantitative mineral abundance information from mineral distribution maps were collected using a Tescan Mira 3 field emission scanning electron microscope (FEGSEM) with Tescan integrated mineralogy analyser (TIMA) at the Australian Resources Research Centre (ARRC, CSIRO Mineral Resources Perth). Different types of samples were analysed, including 25 mm polished rounds, polished thin-sections and ~ 10 cm long polished cores. Further details on the analytical method are provided in ESM 1.

Maia mapper

Two core samples were analysed at CSIRO ARRC, Perth, on the Maia Mapper (Ryan et al. 2018); a laboratory based µXRF mapping system using the Maia Detector Array (Kirkham et al. 2010; Ryan et al. 2014) with a 30 μm spot size and 4 ms dwell time for each pixel. From these large XRF maps, semi-quantified trace element analysis can be extracted using GeoPIXE hotspot analysis as well as determination of locations and size of rare phases, such as zircon.

XFM beamline-synchrotron

Two samples were analysed with the Maia detector on the X-ray fluorescence (XFM) beamline at the Australian Synchrotron to obtain a fine-scale resolution of 2–4 μm. Further details on this approach can be found in Paterson et al. (2011); Ryan et al. (2014); Barnes et al. (2020a), and the full method for data collection in Mao et al. (2019).

High-resolution X-ray computed tomography

One sample from MAD177 from the base of the massive sulfides zone was scanned using high-resolution X-ray computed tomography (HRXCT) at CSIRO ARRC, Perth, to determine the 3D distribution, shape and proportion of sulfides within the sample. Further details on the analytical method are provided in ESM 1. Sulfides were segmented using the methods described in Godel (2013).

Bulk-rock chemistry

One representative sample from each lithological unit (chilled margin, lherzolite, clinopyroxenite, leucogabbronorite and host granitic rock) were analysed for bulk-rock geochemistry. Samples were prepared and analysed by Labwest (Scheme code LITH-02R) using a combination of alkaline fusion and HF/multiacid microwave digestion. Elements were measured by ICP-OES and ICP-MS and FeO was determined titrimetrically. Analyses from the Cathedrals mafic intrusion were compared to regional geochemical data from the Widgiemooltha Dolerite (data from the GSWA WACHEM database; Riganti et al. 2015). The results are provided in ESM2 Table A1.

Mineral chemistry

Point analyses of spinel, arsenides and silicates and trace element map of chromite were collected using a Photonmachines, ATLex 300si-x Excite 193 nm Excimer ArF laser connected to an Agilent 7700 ICP-MS located in the ARRC Geochemistry Laboratory at CSIRO in Perth. The methodology is described in ESM 1 and the results provided in ESM 2 Table A2.

LA-ICP-MS U-Pb dating

The Cathedrals mafic intrusion is Zr-undersaturated and hence does not contain zircons. For this reason, U-Pb dating of apatite was performed to determine the emplacement age of the Ni-hosting mafic intrusion. We utilised TIMA analyses and grain separation to locate apatite crystals. To provide additional constraints, we also conducted U-Pb dating of zircons from granites sampled below and above the mafic intrusion, and from resorbed quartz phenocrysts within the leucogabbronorite section. These are granitic relicts that have been incorporated within the mafic intrusion during its emplacement. In addition to magmatic ages, Pb-loss ages can provide insight into when zircons have been disturbed by thermal or fluid-related events, such as during emplacement of the mafic intrusion in the Cathedrals belt. Zircons from the host-rock granite at the Cathedrals and Investigators prospects, and apatite from the mafic intrusion at the Investigators and Stricklands prospects were analysed. Overall, 11 samples from 6 drill cores were selected for U-Pb dating of zircon, and 7 samples from 5 drill cores were selected for U-Pb dating of apatite. Apatite was dated in-situ from the 25 mm polished rounds. Several sample preparation techniques were used for zircon dating, as described in ESM 1. The results are provided in ESM 2 Tables A3&A4.

Sulfur isotopes

The sulfur isotope composition of pyrrhotite, pentlandite, chalcopyrite and pyrite from sample MAD177 from the base of the massive sulfide zone and a chilled margin were measured in situ by secondary-ion mass spectrometry (SIMS) using the Cameca 1280 Ion Probe housed at the Centre for Microscopy and Microanalysis (CMCA) of the University of Western Australia. Further details on the analytical methods are provided in ESM 1 and the results are in ESM 2 Table A5.

Results

Petrographic and internal differentiation analysis

Lithologies

In the study area, the host-rocks consist of granodiorite to monzogranite, quartz-K-feldspar pegmatite, tholeiitic basalt and amphibolite. The latter two lithologies occur predominantly at the Stricklands prospect at the northern extension of the western belt (Fig. 2c). The cross-cutting Cathedrals mafic intrusion typically comprises, from top to bottom (Fig. 4; ESM 3-Fig. 1): (1) leucogabbronorite that contains resorbed granitic and pegmatitic fragments with resorbed quartz phenocrysts; (2) clinopyroxenite, and (3) lherzolite with increasing proportions of Ni-Cu-Fe sulfides at greater depth, ranging from globular disseminated sulfides (occasionally with globule-bubble pairs), to net-textured sulfides, and massive sulfides. The petrographic characteristics of those units are described below.

Fig. 4

Typical geological log for drill core intersecting the Cathedrals mafic intrusion. The differentiation of the sill from massive sulfides, net-textured sulfides, ultramafic layer, gabbronorite and leucogabbronorite comprising resorbed fragments of granitic and pegmatitic country rock are shown

Upper contact and hybrid zone: leucogabbronorite

The upper contact between the granitic host-rock and the mafic intrusion is typically gradual (ESM 3-Fig. 2a-d). Granitic, pegmatitic and quartz vein fragments are common at the top of the mafic intrusion forming a hybrid zone up to ~ 35 m thick (ESM 3-Fig. 2e). The leucogabbronorite comprises resorbed quartz phenocrysts within a medium-grained matrix of albite-anorthoclase, microcline, quartz, minor clinopyroxene, and in places actinolite. This part of the mafic intrusion is more differentiated with up to 40 vol% plagioclase. TIMA analyses plot in the leucogabbronorite to anorthosite fields (ESM 3-Figs. 1 and 2).

Clinopyroxenite - gabbronorite layer

The transition between this hybrid zone and the lherzolite consists of an olivine gabbronorite and in places, clinopyroxenite (Fig. 4). The clinopyroxenite exhibits euhedral tabular pigeonite crystals (Ca ~ 1–2 wt%) and minor augite, both surrounded with small biotite crystals, and anhedral albite-anorthoclase (ESM 3-Fig. 4). The olivine gabbronorite comprises euhedral olivine phenocrysts that have been partially to fully serpentinised with magnetite in cracks, and interstitial clinopyroxene, pigeonite, plagioclase and minor alkali-feldspar. The lherzolite and olivine gabbronorite form a continuum from mesocumulates with high Mg contents, orthocumulate with intermediate Mg, and gabbronorite with lower Mg.

Cumulate layer

The lower ~ 15 m of the section in hole MAD031 (Fig. 4) comprises an ultramafic cumulate layer, specifically an olivine orthocumulate (lherzolite in Streckeisen terminology). This unit contains a framework of original euhedral olivine grains with an interstitial pyroxene-plagioclase assemblage representing the original parent liquid trapped between the accumulated olivine grains and minor primary (cumulus) chromite ESM 3-Fig. 3. This unit typically contains disseminated sulfides, commonly showing globular textures (Top of Fig. 5a).

Fig. 5

Transition of disseminated globules of sulfides in lherzolite to net-textured sulfides and massive sulfides: (a) Sample photo; (b) TIMA image; (c) High Resolution X-Ray tomography illustrating the 3D distribution of sulfides (massive sulfides in red; interconnected sulfide network in the upper part of the sample in orange; sulfide globules with connected network in purple; finely disseminated sulfide in yellow; see text for further details)

Sulfide droplets

In the lherzolite layer, there is a transition from top to bottom of disseminated sulfide globules, net-textured sulfides and massive sulfides, as observed in MAD031 at ~ 110 m depth (Fig. 5). Ni-Cu sulfide globules are in places associated with amygdales or segregation vesicles, similar to those observed in the Norilsk-Talnakh deposits (Barnes et al. 2019), that are filled with silicate minerals such as plagioclase derived from differentiated trapped liquid. (Fig. 6). These centimetre-sized sulfide globules are commonly subhorizontal and flattened, with the flattened long axis perpendicular to the subvertical core axis (ESM 3-Fig. 4). The sulfide globules contain dominantly pyrrhotite with exsolved pentlandite and an upper chalcopyrite margin. Chromite and magnetite are concentrated at the contact between Ni-Cu-Fe sulfide globules and veins with the lherzolite (Fig. 7). Massive sulfides commonly exhibit a loop texture with pyrrhotite grains surrounded by pentlandite and chalcopyrite (Fig. 4). These loop textures are also observed in other intrusion-related Ni deposits such as Nova-Bollinger and Norilsk ores (Barnes et al. 2020b). Scarce millerite is observed, likely replacing original pentlandite in the disseminated sulfides.

Fig. 6

Droplet-bubble pair in MAD031: (a) TIMA image; (b) Synchrotron Maia XFM image

Fig. 7

Chromite at the contact between lherzolite and Ni-Cu-sulfides: (a) TIMA image of euhedral and skeletal phenocrysts of chromite surrounded by a magnetite rim; (b) Synchrotron Maia XFM image of a Ni-Cu sulfide vein in lherzolite with concentration of chromite (red) at the margins

Bottom contact

The bottom contact between the cross-cutting intrusion and the granite host-rock is typically, albeit not exclusively, sharp. A chilled margin is observed at the bottom of MAD177 and contains disseminated, well rounded, sulfide droplets comprising pyrrhotite, pentlandite, chalcopyrite and in places secondary pyrite (ESM 3-Fig. 5a-b). Similar to the upper contact, the host granite at the footwall has been thermally affected showing features such as resorbed crystals (mostly quartz) at the contact with the mafic intrusion. At the Stricklands prospect, the footwall consists of a Ti-rich amphibolite dominated by clinozoisite and amphibole. The lower contact is gradational as illustrated by the progressive increase in mafic minerals and iron content in the footwall granite towards the mafic intrusion (ESM 3-Fig. 5c-d). Fe-Ni-Cu sulfide veins cross-cut the footwall granite. One vein contains pentlandite crystals with a rosette texture (ESM 3-Fig. 6a-b), and another sulfide vein is cross-cut by a carbonate vein with gersdorffite, galena and sphalerite (ESM 3-Fig. 6c-d).

Within the host granite at the bottom contact with the mafic intrusion, quartz phenocrysts are resorbed, alkali-feldspar content decreases while secondary muscovite is more abundant. Massive sulfides interacted with the footwall granite, as illustrated by emulsion textures near the bottom of MAD177 (ESM 3-Fig. 6e-f). In this zone, relicts of granite within the mafic intrusion consist of a deformed mineral assemblage comprising muscovite, microcline, plagioclase, and clinochlore; the mafic intrusion and associated sulfides are represented by clinochlore (replacing original mafic minerals), pyrrhotite, pentlandite, skeletal chromite, calcite and quartz. Secondary minerals calcite and quartz occur towards the edge of sulfide globules, whereas skeletal chromite occurs within the surrounding clinochlore zone at the margin of sulfide globules (ESM 3-Fig. 6f). Chromite is typically surrounded by magnetite.

3D distribution of sulfides - HRXCT

A 12 cm long sample from the contact between the olivine orthocumulate layer and the massive sulfide zone was analysed to determine the 3D distribution of sulfides. Sulfides were separated into four categories based on their textures and size: Texture A (red in Fig. 5) represents massive sulfide at the bottom with its overlying interconnected network of sulfides; Texture B (orange in Fig. 5) represents interconnected sulfide networks located in the upper portion of the samples; Texture C (purple in Fig, 5) represents sulfide globules with connected networks; and Texture D (yellow in Fig. 5) represents finely disseminated, isolated sulfides with sulfide particle size smaller than 1,000 μm equivalent sphere diameter.

The transition between massive and net-textured sulfides is marked by a change in sulfide abundance with close to 100 vol% in the massive sulfide zone and around 15 vol% in the net-textured zone. The latter unit consists largely of sulfides that are interconnected and surrounding the silicate framework minerals. Sulfide abundance in the net-textured sulfide zone shows a variation from 10 to 25 vol%. Minor droplets (yellow and purple in Fig. 5) occur within this zone (up to 15 vol%). The abundance of these droplets is significantly greater above the net-textured sulfide zone, at 6 to 7.8 cm from the base of this sample. In this interval, the sulfides are no longer interconnected; they are finely disseminated isolated sulfide droplets with particle sizes smaller than 1,000 μm equivalent sphere diameter (yellow in Fig. 5). In the upper part, 8 to 12 cm from the base of the sample, are small zones of interconnected sulfides (e.g. ~8 to 10 cm & ~12 cm; orange in Fig. 5) as well as isolated droplets. These isolated droplets (~ 2–6 mm) are more abundant (specifically at 10 to 11 cm above the base of the sample) than in the rest of the sample.

Mineralogy and mineral chemistry

Spinel

Spinel is a common mineral at the contact between Ni-Cu-Fe sulfides and lherzolite (Fig. 7). It also occurs as disseminated grains within the lherzolite and olivine gabbronorite. Dispersed chromite grains are typically small (mostly < 50 μm) whereas those located on the selvages of sulfide accumulations (e.g. veins) are larger. Similarly, magnetite commonly occurs in proximity to sulfides. Occasionally, sulfide globules are rimmed completely by magnetite (MAD026 in ESM 3 Fig. 4). Magnetite is also found in the massive sulfide samples (Fig. 7a).

Spinel minerals from the Cathedrals mafic intrusion comprise pure magnetite, pure chromite, Al-chromite, ferrian-chromite and minor Cr-magnetite. Cr-bearing spinel in the lherzolite, clinopyroxenite and leucogabbronorite has greater Ni content (mostly > 1,000 ppm Ni) compared to that from the net-textured sulfides (ESM 3- Fig. 7). Cr-bearing spinel in the tholeiitic mafic host-rock at the Stricklands prospect is distinct to most of the spinel in the Cathedrals mafic intrusion with much lower Mg content, low Ni content and higher Zn (ESM 3- Fig. 7). Only one sample from the Cathedrals mafic intrusion has Cr-bearing spinel of similar composition to the tholeiitic host-rock. This sample contains emulsion-like textures comprising deformed sulfide globules with secondary calcite and quartz at their border, an altered mafic mineral assemblage dominated by clinochlore, and relicts of granitic rocks represented by muscovite, microcline, plagioclase and sparse zircons. The spinel grains are thus interpreted to have crystallised from a contaminated magma resulting in their peculiar chemical character (ESM 3- Fig. 6c-d).

Large chromite crystals (mm-size) with a surrounding magnetite rim are observed at the contact between lherzolite and massive sulfide in MAD 126 at 190.3 m depth (Fig. 7a). The largest chromite grains appear within a few cm of the contact where blebs of silicate melts have been incorporated within the massive sulfide zones (green fragments in Fig. 7a). Chromite grains are mostly euhedral, however, chromite located at the contact between the silicate and sulfide melt exhibits a skeletal morphology. Additionally, LA-ICP-MS mapping reveals sector zoning highlighted by the increased concentration of Ti and Mg (Fig. 8a). The spinel is surrounded by a thin magnetite rim enriched in Zn (Fig. 8a).

Fig. 8

Mineral chemistry of the Cathedrals mafic intrusion: (a) Close-up chromite from Fig. 7A. Left to right are Mg, Ti and Zn contents respectively, as determined by LA-ICP-MS mapping; (b) Synchrotron Maia XFM image depicting sector zoning in pyroxene in MAD31-108.9; (c) Nickel vs. forsterite content in olivine; barren continental LIP and barren orogenic belts contours are from Barnes et al. (2023); (d) PGE contents of arsenides

Silicate minerals

Synchrotron Maia maps of the lherzolite in MAD031 reveal sector zoning defined by Cr content in clinopyroxene (Fig. 8b), a feature that signifies rapid crystallisation and is common in small mineralised intrusions (Schoneveld et al. 2020). Two types of clinopyroxenes are observed in the Cathedrals mafic intrusion: pigeonite in lherzolite, clinopyroxenite and leucogabbronorite, and augite in all lithologies. REE chondritic normalised patterns for augite and pigeonite are distinct. Augite grains have flat REE patterns above chondritic values, whereas those for pigeonite are characterised by small positive REE slopes from La to Lu with LREE below chondritic values and HREE above chondritic values (ESM 3-Fig. 8). Augite from the net-textured zone is the most depleted in Ni with values < 370 ppm (ESM 3-Fig. 8).

Olivine grains are typically altered to serpentine and/or talc + carbonate, with few fresh cores remaining. Normally, chondrite normalised REE patterns for olivine display a significant depletion in the LREE due to their incompatibility in olivine. Most REE patterns of olivine measured in this study lack the negative LREE slope expected for olivine, suggesting these grains have been altered (ESM 3-Fig. 8). This is supported by the high levels of Al (740 to > 3,000 ppm Al) measured in olivine with La normalised to chondritic values > 0.1. Forsterite (Fo) contents of olivine range from 80 to 90 as determined by LA-ICP-MS, consistent with Fo values acquired by EPM averaging at 86.7 +/-2.9. The Fo content of olivine increases from the top clinopyroxenite layer (average Fo85.8) to the lherzolite layer (average Fo88) and to the net-textured sulfide zone below (average Fo89.6) (Fig. 8c). Nickel contents analysed by LA-ICP-MS range from 2,000 to 3,400 ppm, consistent with average values of 2,625 +/- 303 ppm determined by EPM. The Ni contents in relation to Fo in these olivine grains are fairly typical of olivine in intrusions in continental LIPs, falling at or just beyond the high-Fo end of the 80th percentile field defined by Barnes et al. (2023) (Fig. 8c) and showing no evidence of Ni depletion.

Plagioclase from the host granite is albitic (7 to 7.5 wt% Na and < 3 wt% Ca) compared to the mafic intrusion where it ranges from oligoclase to labradorite (6–12 wt% Ca and Na < 5.5 wt%). All grains exhibit a typical positive Eu anomaly. Plagioclase from the net-textured sulfide zone is more enriched in HREE with chondrite normalised values up to 10, whereas in other lithologies it remains mostly below 1 (ESM 3-Fig. 8). In the sulfide zone plagioclase is also more enriched in Ni, with median contents of 220 ppm. In contrast, in other lithologies the median Ni contents of plagioclase is < 6 ppm.

Sulfarsenides

Arsenic-rich minerals are scarce and found in small amounts along the selvages of late-stage carbonate veins at the intersection with Fe-Ni-Cu sulfides (ESM 3-Fig. 6c-d). They are primarily sulfarsenides, forming anhedral grains that crystallised at the margins of quartz grains. They contain numerous galena inclusions. The most abundant phase is gersdorffite (NiAsS) with lesser cobaltite (CoAsS) in proximity to the sulfide vein. Trace element analyses show high Au (mean ~ 0.7 ppm for gersdorffite; n = 13; and 4 ppm for colbaltite; n = 2) and Ag (mean ~ 6 ppm for gersdorffite; n = 13; and 27 ppm for colbaltite; n = 2) with moderate Pd mean ~ 10 ppm for gersdorffite; n = 13; and 104 ppm for colbaltite; n = 2 (Fig. 8d). Ir-group PGEs (IPGEs: Ru, Os, Ir), Rh and Pt are mostly below the detection limit. Similar PGE signatures are observed for arsenides and sulfarsenides found in hydrothermal environments, such as the hydrothermally altered areas at Miitel (Le Vaillant et al. 2015) and at Sarah’s Find (Le Vaillant et al. 2016) in the Yilgarn Craton (Fig. 8d).

U-Pb dating of the cathedrals mafic intrusion and host-rock granite

Zircon grains from mineral separates of sample MAD177-198 m from the host-rock granite range in size from ~ 50 to ~ 300 μm, are euhedral, brown in transmitted light and have numerous fractures. Reflected light microscopy shows unusual inclusions of chalcopyrite (ESM 3-Fig. 9a). Zircon grains extracted from MAD177-198 also contain fluorite and calcite inclusions (ESM 3-Fig. 9b), as well as more typical inclusions of apatite and rutile. Cathodoluminescence images reveal common oscillatory zoning in rims, whereas some zircon grains exhibit a darker resorbed core (ESM 3-Fig. 9c-d). The mineral separates of sample MAD006-120 from the host-rock granite yielded fewer and smaller zircon grains than those from MAD177-198; their size rarely exceeds 200 μm. MAD006-120 zircon grains are transparent to brown, the latter displaying significant fractures. Their cathodoluminescence response is similar to MAD177-198 zircons (ESM 3-Fig. 9).

A summary of the results is provided in Table 1 and illustrated in ESM 3-Fig. 10. The crystallisation age of the host-rock granite, obtained from the zircon grains, ranges from 2741 Ma (MAD31-118.4) to 2549 Ma (MAD177-198), with most samples yielding an age of c. 2660 Ma (Table 1). These results are consistent with regional geochronology datasets that show abundant magmatic activity at c. 2660 Ma in the Kalgoorlie Terrane (Mole et al. 2019; Zibra et al. 2020). These new data provide maximum age constraints for the emplacement of the Cathedrals mafic intrusion.

Table 1 Summary of U-Pb dating of zircons

The results also indicate a dominant inherited zircon population at c. 3.1–2.9 Ga. Most zircon grains have experienced partial Pb loss. The data reveal three Pb loss events at c. 1100 Ma, c. 600 Ma and present-day (Table 1; ESM 3-Fig. 10). The c. 1100 Ma Pb loss age may correspond to the c. 1200 Ma Marnda Moorn LIP or the c. 1075 Ma Warakurna LIP. A sill dated at 1070 ± 18 Ma occurs north of the project area (Fig. 2a-b) and it is likely that other coeval intrusions occur nearby as are common throughout the northwestern Yilgarn Craton (Wingate et al. 2008; Wingate 2017). The c. 600 Ma Pb-loss age has previously been observed in the Eastern Goldfields Superterrane and interpreted as spatially associated with the undated, northwesterly trending Cosmo Newbery Dyke Suite (Kirkland et al. 2017).

Apatite grains in samples from the Cathedrals mafic intrusion are scarce and small in (< 100 μm) and occur both within the silicate melt and in disseminated sulfide globules. The regression of 21 apatite analyses anchored to a measured common Pb value yielded an age of c. 2336 ± 64 Ma (Fig. 9), interpreted as the magmatic crystallisation age of the mafic intrusion. This is consistent with the age of the Widgiemooltha Dolerite dated at 2408+/-3 Ma and 2401+/-1 Ma (Pisarevsky et al. 2015; Wingate 2017).

Fig. 9

Terra-Wasserburg diagram of apatite from the Cathedrals mafic intrusion anchored to a measured common Pb value of 1.068 +/- 0.094 using the new model-3 anchored regression algorithm in Isoplot-R (Vermeesch 2024)

Bulk-rock chemistry

REE spider diagrams normalised to chondritic values of Taylor and McLennan (1985) and Evensen et al. (1978) illustrate an overall flat REE pattern with slight enrichment in LREE (Fig. 10a). REE abundances in the lherzolite are much lower than the rest of the mafic intrusion, reflecting the cumulate character of these rocks, and similar patterns are found in the Celebration Dolerite Member and the Jimberlana Norite Member (Fig. 10a; data from the GSWA WACHEM database; Riganti et al. 2015). The leucogabbronorite has stronger LREE enrichment in comparison to the chilled margin, lherzolite and clinopyroxenite and its REE pattern lies in between the host-rock granite and chilled margin, consistent with the incorporation of host-rock granite observed in this unit. This signature is comparable to the Widgiemooltha Dolerite (open blue symbols in Fig. 10a).

Fig. 10

Plots showing whole-rock geochemical data from this study (Cathedrals mafic intrusion) and regional geochemical data from Widgiemooltha Dolerite intrusions (Geological Survey of Western Australia; Geoview): a. REE spider diagrams normalised to chondritic values of Taylor and McLennan (1985) modified from Evensen et al. (1978); b. Ni vs. MgO baseline data using assay data filtered for S < 1,000 ppm (grey rectangles; data provided by St George Mining Ltd) with its polynomial fit (black dashed line). Legend as per A; c. Th/Yb vs. Nb/Yb plot highlighting the crustal contamination of the magmas; d. Cu vs. Zr contents revealing the occurrence of disseminated sulfides in the lherzolite and the chilled margin. The clinopyroxenite plots on the mantle ratio line, while the remaining data plot below this line suggesting that sulfides may have been extracted from those magmas (Barnes 2023)

The sample interpreted as potentially having a composition close to the parental magma composition is from the bottom chilled margin indicating an initial MgO content of 12.9 wt% (normalised on an anhydrous basis). Initial nickel content of the parental magma cannot be easily established from this sample due to the presence of disseminated sulfides, including pentlandite. Assay data provided by St George Mining Ltd indicate a polynomial relationship between Ni and MgO (for samples with S < 1,000 ppm; n = 5,355); suggesting that magma with 12.9 wt% MgO had about 300 to 500 ppm Ni (Fig. 10b).

The involvement of upper crustal material in the Cathedrals mafic intrusion and also in the Widgiemooltha Dolerite (as suggested by the REE patterns) is evident from elevated Th/Yb and Nb/Yb ratios above those of primitive mantle and aligning along the PM-UCC mix line of Barnes (2023). Their composition plots within the field of basalts from Continental Large Igneous Provinces (CLIP). The clinopyroxenite and chilled margin of the Cathedrals mafic intrusion, the gabbronorite of the Jimberlana Norite Member and the gabbro of the Celebration Dolerite Member have the least crustal influence in their signature (lower Nb/Yb and Th/Yb on this mix line). The latter two plot under the Cu-Zr mantle line ratio suggesting that sulfides may have been extracted from those magmas (Barnes 2023). This is consistent with the occurrence of magmatic nickel sulfides at Cowarna Rocks within the Celebration Dyke and within the Jimberlana Norite Member. The chilled margin and the lherzolite contain magmatic Ni sulfides (as shown by the higher contents in Ni and Cu in comparison to the other samples) and thus they plot above the Cu-Zr mantle ratio line (Fig. 10b, d). In contrast, the clinopyroxenite plots on the Cu-Zr mantle line, indicating neither accumulation nor removal of sulfides. The leucogabbronorite plots below the Cu-Zr mantle ratio line, due to either (1) the incorporation of resorbed fragments of granitic rocks and/or (2) extraction of sulfides.

Sulfur isotopes

Sulfur isotopes measured in chalcopyrite, pyrrhotite, pentlandite and pyrite from sulfide globules in the chilled margin and from disseminated and net-textured sulfides have δ³⁴S values ranging from − 2‰ to + 2‰, and plot within the mantle range (Sakai et al. 1984; Chaussidon et al. 1991; Labidi et al. 2014) (Fig. 11). Isotopic data from sulfide globules in the chilled margin are the most positive, with δ³⁴S mean values of -0.5‰, 0‰, 0.5‰ and 0.75‰ for pyrite, pentlandite, pyrrhotite and chalcopyrite, respectively. In contrast, the net-textured sulfides have mean δ³⁴S ranging between − 0.5 and − 0.7‰, and the disseminated sulfide globules in the lherzolite are < -0.9‰ (Fig. 11a). The bulk δ³⁴S for each unit was determined using quantitative TIMA mineralogy confirming the trend towards more negative δ³⁴S values from the chilled margin towards the disseminated sulfide globules. All sulfur isotope data align on the Mass Dependent Fractionation (MDF) array, without any significant deviations (Fig. 11b). Chalcopyrite in the chilled margin exhibits the most positive values reaching a mean δ³⁴S value of + 0.75‰. This is followed by pyrrhotite, pentlandite and pyrite with respective mean δ³⁴S values of + 0.5‰, 0‰, and − 0.3‰ in the chilled margin (Fig. 11a).

Fig. 11

Sulfur isotope data from sulfides of the Cathedrals mafic intrusion: (a) Boxplots of δ³⁴S for chalcopyrite, pyrrhotite, pentlandite, and pyrite organised by units. The blue line indicates the bulk δ³⁴S derived from quantitative TIMA mineralogy. Note the overall δ³⁴S decrease from the chilled margin to the net-textured and disseminated sulfides; (b) δ³³S vs. δ³⁴S illustrating the alignment of sulfur isotope data from the sulfides of the Cathedrals mafic intrusion with the Mass Dependent Fractionation (MDF) array

Discussion

Parental magma, emplacement, and geometry of the mafic intrusive system

Parental magma

The parental magma of the Cathedrals mafic intrusion contained around 13 wt% MgO and 9 wt% FeO as constrained from geochemical analysis of the chilled margin (ESM 2 Table A1). Olivine crystallising in equilibrium with this magma composition should have Fo contents of 88.3 (using Kd_Mg−Fe of 0.34 from Matzen et al. (2011)), values that are in agreement with data collected in this study (e.g., EPM data of 86.7 +/-2.9). The Ni content of the parental magma cannot be reliably estimated from the chilled margin due to the occurrence of disseminated magmatic sulfides. Instead, we utilise the relationship between Ni and MgO with less than 1,000 ppm S based on whole-rock assay. Knowing the magnesian composition of the parental magma (12.9 wt% MgO), this correlation provides an estimate of the initial Ni content, ranging from ~ 300 to 500 ppm (Fig. 10b). On this basis, sulfides with the average composition of the massive ores (6 wt% Ni) would have formed at R values (silicate-sulfide mass ratio) of around 150–200. This suggests the olivine should have Ni contents of ~ 500 ppm, well below the observed values of ~ 2,600 ppm, assuming oxygen fugacity at QFM. Twenty-five to 30 g of olivine per 100 g of silicate melt would be required in the initial assemblage to explain the highest Ni tenors of ~ 16% observed in the massive sulfides, using the calculation of Barnes et al. (2013). However, the calculated Ni content of olivine under these conditions remains too low at ~ 1,500 ppm. The discrepancy with the Ni content is discussed below in the Mineral indicators section.

In the Cathedrals mafic intrusion, centimetre-sized sulfide globules above the net-textured and massive sulfide zones are commonly flattened (ESM 3-Fig. 4). In addition, these sulfide globules commonly exhibit internal differentiation (Fig. 6a) with a mineral assemblage of pyrrhotite and exsolved pentlandite interpreted as the subsolidus breakdown of monosulfide solid solution (MSS) and an upper chalcopyrite margin derived from original intermediate solid solution (ISS) (Craig and Kullerud 1969). The latter is interpreted as the result of fractional crystallisation of the Ni-Cu-Fe sulfide liquid by enrichment of Cu in the residual liquid due to MSS crystal fractionation (Barnes et al. 2006), which is the major constituent of more differentiated sulfide droplets. Some Ni-Cu sulfide globules have an upper silicate cap that consists of an amygdale or segregation vesicle filled with more differentiated silicate minerals such as plagioclase (Fig. 6). These are interpreted as bubbles carrying and transporting Ni-Cu sulfides in the melt (refer to Barnes et al. 2019 for a discussion on the ore-forming process). These droplet-bubble pairs have been noted in other economic sulfide deposits such as the Norilsk-Talnakh camp in Siberia (Le Vaillant et al. 2017; Barnes et al. 2019) and the komatiite-hosted Black Swan deposit in Western Australia (Dowling et al. 2004; Hill et al. 2004). The differentiated part of the sulfide globule and the occurrence of a silicate cap above the sulfide droplet (Fig. 6) indicate the paleo-up direction. The flattening long axis of the globules is perpendicular to the subvertical core axis, indicating the intrusion is subhorizontal. Thus, the geometry and paleo-up direction indicate that the Cathedrals mafic intrusion was emplaced as a sill that is right-way up, rather than a dike.

Another feature supporting the emplacement of the Cathedrals mafic intrusion as a sill is its internal differentiation. The sill is typically differentiated into a lower ultramafic (olivine) cumulate layer, representing downward accumulation (probably by settling) of dense olivine crystals, overlain by a dolerite unit containing variable proportions of assimilated granitoid, representing the liquid-rich upper portion of the sill contaminated by roof melting. These granitoid clasts are buoyant in mafic magma and would therefore tend to accumulate at the top of the sill at the same time as the olivine is accumulating at the bottom. The extent of melting and assimilation depends on the time available before the solidification of the sill. Additionally, the transition from top to bottom of disseminated sulfide globules, net-textured-sulfides and massive sulfides (Fig. 5) reveal the migration and settling of sulfide droplets and accumulation into a net-textured sulfide and massive sulfide zone through sulfide percolation and growth of the sulfide network by coalescence (Barnes et al. 2017). The direction of this settling also confirms the paleo-up direction of this sill.

Possible emplacement styles include a saucer shape morphology (Fig. 12a) or a set of en-echelon segments within a sill interconnected by inclined sheets (Fig. 12b) (Schofield et al. 2012; Magee et al. 2016). Saucer shape sills, while mostly occurring in sedimentary host-rocks (e.g., Magee et al. 2016), also occur within crystalline or granitic basement (Brown and Kim 2020) such as in the Fennoscandian shield (Buntin et al. 2019), the Karaj Dam basement sill in Iran (Maghdour-Mashhour and Shabani 2017), and the Warakurna sills in the Yilgarn Craton. In the latter, seismic reflection profiles show saucer shape morphologies that were attributed to the Warakurna sills (Wyche et al. 2014). The alternative scenario of en-echelon segments interconnected by inclined sheets, also called intrusive steps and bridge structures (Fig. 12a), typically starts with the intrusion of thin isolated magma segments that inflate and amalgamate with sustained magmatic input (Magee et al. 2016). Evidence of magmatic inflation is observed in MAD177 and MAD181 (ESM 3-Fig. 11) with the occurrence of granitic xenoliths within the lherzolite layer in MAD181, and a ~ 2 m lherzolite unit below the massive sulfide zone in MAD177. These intervals are interpreted as rafts from earlier magmatic input that were buried towards the bottom as inflation progressed. The Cathedrals mafic intrusion is interpreted as a sill complex fed by deeply-sourced magmas. It is quite likely that the adjacent Ida Fault, a deep crustal boundary which separates the Eastern Goldfields Superterrane from the Younami Terrane facilitated magma flow into its hanging wall and along the Cathedral Fault (Figs. 1 and 2). The Norilsk-Kharealakh Fault is a similar long-lived deep crustal fault system (Naldrett 1992), that locally acted as a magma conduit.

Fig. 12

Sill emplacement: (a) saucer shape morphology; (b) set of en-echelon segments within a sill interconnected by inclined sheets

Sulfur source

While it is well accepted that the assimilation of sulfur-bearing rocks is the most efficient process to develop a sulfide liquid within mafic magmas, sulfur isotope data from the sulfides in the Cathedrals mafic intrusion do not show evidence of assimilation of crustally-derived sulfur. The Cathedrals mafic intrusion was emplaced at ~ 2400 Ma, close to the Great Oxidation Event (GOE) revised at 2.33 Ga (Luo et al. 2016). Therefore, we consider the assimilation of sulfur-bearing rocks formed either before and/or after the GOE. Typically, Archean sediments, those formed prior to the GOE have strong Mass-independent fractionation (MIF) signatures and a narrow range of δ³⁴S values, whereas post-GOE sediments lack significant MIF signatures and have a broad δ³⁴S range (Johnston 2011; Fakhraee et al. 2018; Killingsworth et al. 2019). The sulfur isotope data collected from sulfides from the Cathedrals mafic intrusion vary within a narrow δ³⁴S range (-2 to + 2‰) and are aligned on the Mass Dependent Fractionation (MDF) array without any significant deviations attributable to photochemical MIF (LaFlamme et al. 2018) (Fig. 11). Consequently, the sulfur isotope data do not provide evidence of assimilation of either Archean or post-GOE sediments in the Cathedrals mafic intrusion due to the lack of photochemically-derived MIF-signatures and δ³⁴S variations beyond the mantle range (Fig. 11).

An additional factor to consider is the tendency of magmatic sulfides to become partially reset due to dilution by mantle S from the carrier silicate melt (Lesher and Burnham 2001). At the estimated R factor value of around 150 derived above from the massive sulfide Ni tenor and the chilled margin composition, assuming a basaltic melt with up to 500 ppm S at + 0.07 δ³⁴S and complete equilibration between the sulfide and silicate melt, the relative magnitude of the dilution effect on δ³⁴S would be no more than 20%, well within the observed range of variation (See Eq. 16 in Lesher and Burnham 2001).

The sulfur isotope data available neither confirm nor exclude the contribution of pre- or post-GOE sediments to the formation of the sulfide liquid in the Cathedrals mafic intrusion. While the S-rich sediments of the 2690–2665 Ma Black Flag Group (Krapez 1997; Krapez et al. 2000), located close to the Cathedrals mafic intrusion, are not a good candidate for the sulfur source due to their distinctively positive MIF-signature (Chen et al. 2015), the mafic melt may have interacted with other sulfur sources elsewhere. This overall scenario mirrors the isotopic features of the komatiite-hosted Long Victor shoot of Kambalda, where basal massive Ni sulfides have δ³⁴S values within the mantle-value range, and no MIF signatures (Bekker et al. 2009). At Long Victor, these features have been interpreted to reflect either a magmatic sulfur source or assimilation of crustal materials with weak MIF signatures that were diluted in a dynamic magmatic system. Similar conclusions can be drawn for the Cathedrals mafic intrusion given that existing evidence does not permit confidence to establish its origin. Additional sulfur isotope data from sulfides from the komatiitic sequence of the Ularring greenstone belt (e.g. the Mount Alexander prospect of St George Mining Ltd) may shed light on the sulfur source of the Cathedrals mafic intrusion.

Isotopic variations between samples introduce an additional level of complexity when attempting to trace the origin of sulfur within the Cathedrals mafic intrusion. The trend towards more negative δ³⁴S values (~ 1‰ isotopic shift) from the earlier crystallised sulfide in the chilled margin to the late crystallised disseminated sulfide globules in the lherzolite likely developed during emplacement as a result of magmatic degassing, thereby modifying the isotopic signature of the magmatic sulfides. The presence of segregation vesicles associated with sulfide blebs suggests that gas bubbles in the mafic melt interacted with the sulfide liquid both mechanically and chemically. This interaction likely facilitated the transport of sulfide blebs through the magma conduits as well as the formation of large sulfide accumulations through coalescence (Mungall et al. 2015; Iacono-Marziano et al. 2022). The association of gas vesicles and sulfide droplets suggests that decreasing δ³⁴S values from globules to disseminated sulfides may be caused by local desulfurisation. In this scenario, the exsolution of volatiles from the sulfide melt was likely triggered by sudden pressure drops in the magma conduits (Kavanagh et al. 2015). As degassing effects are more intense on small sulfide aggregates, it is plausible that the decreasing δ³⁴S values from globules to disseminated sulfides reflect the loss of isotopically-heavy SO₂-rich volatiles from the sulfide liquid (Marini et al. 2011; Caruso et al. 2017).

Mineral indicators for Ni mineralisation in mafic-ultramafic intrusions

Mineral chemistries can record magmatic histories such that these mineral compositions can reflect ore forming processes, whether it be sulfide saturation, high R-factors or sulfide accumulation. Several geochemical characteristics of minerals have been proposed to distinguish barren and magmatic Ni-sulfide mineralisation, such as the Ni content of olivine in mafic and ultramafic rocks (e.g., Häkli 1971; Naldrett et al. 1984; Barnes et al. 2004; Barnes et al. 2023), intra-grain Cr-zonation in pyroxene (Schoneveld et al. 2020), compositions of magnetite (e.g. Boutroy et al. 2014; Dare et al. 2014; Ward et al. 2018), and PGE contents in arsenides (e.g., Ahmed et al. 2009; Le Vaillant et al. 2018). These minerals can be measured from within the intrusion or used as ex-situ indicators in cover to give an understanding of the prospectivity of the region or intrusion for nickel sulfide exploration. In this study we investigated some of these minerals within the Cathedrals mafic intrusion. The indicator mineral chemistries were developed in mineralised intrusions that are larger than a sill so it is important to investigate whether they remain useful in these smaller systems.

The calculated composition of olivine that would be in equilibrium with the parent magma and the measured sulfide tenor would be olivine with a Fo value of ~ 88 and Ni ~ 500 ppm. The measured Ni content in olivine ranges from 2,000 to 3,400 ppm showing no evidence for the expected depletion of Ni due to competition with sulfide. These compositions largely overlap the field for unmineralised intrusions in continental LIP settings (Fig. 8c). The olivine is significantly out of equilibrium with the massive sulfides at the R factors indicated by the ore tenors, implying that the lherzolite cumulates might have been emplaced in the inflating sill after the early injection and settling of the sulfides. Ni contents in augite and Cr-bearing spinel are lower in the net-textured sulfide zone compared to the overlying lherzolite and pyroxenite layers. This is explained by the preferential incorporation of Ni into the sulfide melt component, but this feature is only noted in immediate proximity to known sulfides and is not an intrusion-wide signature.

Clinopyroxene is sector zoned with some of the crystal axes more enriched in Cr (Fig. 8b). This signifies rapid growth of clinopyroxene (Ubide et al. 2019), which can occur due to a number of different factors including a drop in temperature or pressure or an addition of silica into the magma, such as from melting of the surrounding wall-rock. Schoneveld et al. (2020) suggest that sector zoning, when paired with other types of intra-grain Cr zoning (e.g. abrupt or oscillatory changes in the Cr content) can signify high flux magma systems and therefore may indicate the potential for higher R-factors leading to sulfide upgrade. Within the Cathedrals mafic intrusion samples, clinopyroxenes are sector zoned with some areas more enriched in Cr (Fig. 8b), however no other types of intra-grain Cr zoning was observed.

We used arsenide chemistries to track potential hydrothermal remobilisation of Ni, Cu, Co, PGEs from a target sulfide accumulation in the vicinity (Le Vaillant et al. 2018); such remobilisation can produce a geochemical halo around the magmatic sulfide accumulation (Le Vaillant et al. 2015, 2016). Ni, Cu, Co, Au, Pd and to a lesser extent Pt are mobile in hydrothermal fluids (especially when there is S in the system), whereas the Ir-group PGEs (IPGEs: Ru, Os, Ir) are not (Barnes and Liu 2012). Therefore, arsenides formed after interacting with magmatic sulfides (which contain Ni and PGEs) will also contain Ni, Co and trace amounts of PGEs, especially Pd and sometimes Pt. Arsenides throughout the Cathedrals mafic intrusion are mostly gersdorffite and cobaltite (contain large amounts of Ni and Co) as well as small amounts of Pd, but very low IPGEs (Fig. 8d). These compositions are in agreement with the interpretation of these arsenides having crystallised from As-rich hydrothermal fluids that had interacted with magmatic nickel sulfides. However, these arsenides were only observed within the mafic intrusion and further work is required to establish the presence of a geochemical halo and occurrence of sulfarsenides in the surrounding country rocks.

It appears that for the Cathedrals mafic intrusion, the chemistry and trace element zoning of the silicate phases (olivine and pyroxene) were not indicative of the presence of sulfides within the intrusion, possibly because the silicate melt present above the massive sulfides is not the melt from which the sulfides were extracted, but rather the melt that would have flown in and crystallised there post-sulfide emplacement. The chemistry of the arsenides present within the intrusion however do indicate presence of magmatic nickel sulfides in the vicinity. To improve the use of indicator minerals for the exploration of mafic-hosted nickel sulfide deposits, a larger compilation of mineral chemistries from small chonolith type intrusions as comparison would be useful. In parallel, collection of lithogeochemistry and mineral chemistries from larger mafic ultramafic bodies that could be interpreted as ‘feeders’ or at least as being part of the same magmatic complex could help evaluate prospectivity.

A new c. 2400 Ma sill-hosted magmatic nickel sulfide deposit

Apatite dated in this study yielded an interpreted crystallisation age of 2336 ± 64 Ma (Fig. 9), which we interpret as Widgiemooltha Dolerite (Wingate 2017). This is supported by the same general east-northeast orientation of the Cathedrals mafic intrusion as the Widgiemooltha Dolerite dikes and its thickness and mafic to ultramafic composition (Figs. 1c and 2a-b). The Widgiemooltha Dolerite dikes exhibit both positive and negative magnetic polarities with the older 2408 ± 3 Ma positive group mostly trending about 075° and the younger, 2401 ± 1 Ma negative group mostly trending about 085° (Pisarevsky et al. 2015; Wingate 2017). These dikes tend to have strong, linear aeromagnetic traces and the Cathedrals belt appears to line up with a series of positive Widgiemooltha Dolerite dikes to the east (Fig. 2B).

Other Ni-occurrences of this age occur in Western Australia. At Cowarna Rocks, 100 km east of Kalgoorlie within the 150 km long and 0.5 to 1 km wide Celebration Dyke, magmatic nickel disseminated sulfide droplets and massive segregations occur within a magnesian tholeiitic basaltic magma intruded into granodiorite (Fig. 1c; Purvis and Moeskops 1981). The ~ 200 km long and 1 to 2.5 km wide Jimberlana Dyke near Norseman (Fig. 1c; Mazzucchelli and Robbins 1973) is considered as a small analogue of the over 500 km long, 2–4 km wide Great Dyke of Zimbabwe (Campbell et al. 1970). The Jimberlana Dyke was dated at 2411 ± 55 Ma using a Sm-Nd whole-rock isochron and although younger, is petrologically similar (Fletcher et al. 1987). Soil samples from above the Jimberlana Dyke contain Ni and Cu anomalies and marginal norite contains minor disseminated pyrrhotite, pyrite, chalcopyrite and pentlandite (Campbell et al. 1970; Mazzucchelli and Robbins 1973; Travis and Knight 1975), but no potentially economic mineralization has yet been located in the Jimberlana intrusion despite multiple exploration programs. The recognition of magmatic Ni-sulfides in the Widgiemooltha Dolerite suggests the 2.4 Ga magmatic event may be more prospective than previously thought, particularly when considered in conjunction with orthogonal host greenstone belts and deep crustal structures.

Conclusions

The Cathedrals mafic intrusion was emplaced as a sill at 2336 ± 64 Ma and is considered part of the Widgiemooltha Dolerite dike swarm. This conclusion is supported by U-Pb dating of apatite, the similar east-northeast orientation of the intrusion and its lithological and mineralisation characteristics. Flattening and differentiation of sulfide globules provide evidence of the subhorizontal orientation of the sill and right way-up direction. The Cathedrals mafic intrusion is vertically differentiated, from bottom to top as: massive sulfides, net-textured sulfides, disseminated sulfides, lherzolite, clinopyroxenite, and variably xenolith-rich leucogabbronorite. The leucogabbronorite formed by contamination of the mafic intrusion with the granitic and pegmatitic host-rock as suggested by the common occurrence of granitic, pegmatitic and quartz vein fragments. Sulfur isotopes have mantle-like signatures with no mass independent fractionation. Magmatic degassing may explain the trend towards more negative δ³⁴S values observed in sulfides within the basal chilled margin towards net-textured sulfides and disseminated sulfides. The mineral chemistries suggest that the sulfide and cumulate crystals do not have a long history of interaction, i.e. the R-factor upgrading occurred elsewhere from the final emplacement location. This causes the chemical signatures of the primary magmatic indicator minerals (i.e. olivine and pyroxene in this case), that have proven effective indicators of prospectivity in larger intrusions, not to be effective here. Secondary minerals such as arsenides still seem to provide interesting insights into the presence of sulfides. New indicator minerals may need to be developed to explore for these smaller, sill-type intrusion hosted magmatic sulfide occurrences. The recognition of magmatic Ni-sulfides in this intrusion suggests the 2.4 Ga Widgiemooltha Dolerite dike swarm event may be a target for further Ni-sulfide discoveries.