Introduction

Conduit-style mafic–ultramafic intrusions are known to host magmatic Ni-Cu sulfide deposits (Ripley and Li 2011; Barnes et al. 2017), whereas intrusions with less magma-throughput (sometimes referred to as “blind” or “closed” intrusions) are considered less prospective due to reduced mechanisms for sulfide liquid generation and metal enrichment. The chemical zonation of Cr and Ti within cumulus pyroxene in such intrusions has been suggested as an indicator for conduit-style intrusion systems and therefore an indicator of enhanced prospectivity for magmatic sulfide deposits (Schoneveld et al. 2020).

Intragrain zoning of Cr within pyroxene was observed in a number of economic magmatic sulfide deposits and was not observed in the barren intrusions of similar age, size, and location (Schoneveld et al. 2020); however, this investigation only considered a binary classification into sulfide-bearing and sulfide-absent (i.e. “barren”) intrusions (with the caveat that false negatives may exist in the barren category).

Pyroxene zoning types

There were four styles of pyroxene zoning that were recognised in the previous study of Schoneveld et al. (2020) (Fig. 1). Normal zoning (Fig. 1a) describes the typical changes in the chemistry of the pyroxene during crystallisation and solidification of the igneous body. This is characterised by a smooth reduction in compatible elements such as chromium, and a corresponding increase in incompatible elements such as titanium. Sector zoning (Fig. 1b) is characterised by an hour-glass or bow-tie shaped variation in chemistry (Fig. 1 B1) that marks the varying partition coefficients of the compatible elements on the different crystal axis of the pyroxene crystal. However, depending on how the crystal is cut, this zonation may appear as a false reverse type zoning (Fig. 1 B2) (i.e. an increase in the compatible element concentration at the rim). Sector zoning likely occurs during rapid crystallisation. Abrupt zonation (Fig. 1c) is characterised by a sudden change in the chemistry of the pyroxene grain, usually high concentrations of compatible elements in the core of the grain that reduces to near zero toward the rim. The final observed zonation type is oscillatory zoning (Fig. 1d), which is characterised by cyclic changes in the elemental concentration. Complex zoning was the term given to any combination of sector, abrupt and oscillatory zoning in a single grain.

Fig. 1
figure 1

Types of Cr zoning in pyroxene (modified from Schoneveld et al. 2020). a Normal zoning; diffuse decrease of the compatible elements such as Cr toward the rims, due to normal crystal growth in an evolving magma. Incompatible elements such as Ti would increase toward the rims. b Sector zoning, due to rapid crystallisation of pyroxene. B1 false reverse zoning due to sectioning through a sector zoned crystal. B2 typical hour-glass shape given by sector zoned crystals. c Abrupt zoning, characterised by rapid change in Cr content within the crystal. d Oscillatory zoning, small cyclic changes in pyroxene chemistry

Full size image

Two important, unresolved questions from the previous study were 1) the extent to which intragrain pyroxene zoning extends away from ore within mineralised intrusions; and 2) the extent to which zonation is recognisable through the imprint of retrograde metamorphism, which commonly results in pseudomorphism of pyroxenes by amphiboles as a result of simple hydration. Understanding the robustness of this prospectivity indicator against this common form of alteration and how far we can expect this zonation to extend beyond the sulfide mineralisation will help to define the usefulness of this indicator for exploration.

To answer these questions, the extent of minor-element pyroxene zoning was examined throughout two intrusive complexes that host magmatic sulfide ore bodies: Nova-Bollinger, Australia, and Kevitsa, Finland.

Geological settings and sampling strategy

Nova-Bollinger

The Nova-Bollinger ore bodies are located within the Albany-Fraser orogen, about 160 km east-northeast of Norseman, Western Australia, and are currently owned and mined by IGO (Fig. 2a). Resource estimates for the deposit are 13.1 Mt at 2.0 wt % Ni, 0.8 wt % Cu, and 0.07 wt % Co (Independence Group, 2019). The Nova-Bollinger orebodies and intrusions have been described in detail in several recent publications (Barnes et al. 2020b; Taranovic et al. 2022b; Torres-Rodriguez et al. 2021) and are only briefly covered here.

Fig. 2
figure 2

a Surface projection of Nova and Bollinger ore bodies and host intrusions within the Albany Fraser Orogen (AFO) with indicated cross section line, modified from Taranovic et al. (2022a) and Barnes et al. (2020a, b). b traces of the sampled drill holes projected onto a simplified composite long section modified from Taranovic et al. (2022a). (Note that most of these drill holes do not lie precisely within the plane of the long section). Zoned pyroxene types observed are described in the results

Full size image

The Nova-Bollinger deposit is made up of two connected intrusions; known as the Lower Intrusion, which hosts the orebodies, and the Upper Intrusion, with no economic mineralisation (Taranovic et al. 2022b). The Lower Intrusion is connected to the Upper Intrusion through a contact zone referred to as the C5 unit (Fig. 2b) which contains minor sulfide matrix breccia veins towards the base and disseminated sulfides throughout.

The Lower Intrusion is characterised by lherzolite/websterite/norite orthocumulates. It contains approximately 50 m of disseminated, net-textured and massive sulfide at the base, with sulfide matrix breccia veins extending up to 100 m into the footwall paragneisses beneath the intrusion (Barnes et al. 2020b). Nova (to the west) and Bollinger (to the east) are two adjacent but spatially distinct orebodies that are both hosted within the Lower Intrusion and their immediately subjacent country rocks. Most of the samples from Nova contain between 5 and 8 wt % Ni and between 0.5 and 10 wt % Cu, compared with 4 to 7% Ni and 0.5 to 8% Cu at Bollinger. Differentiated veins (i.e. pentlandite in the wide part of the vein with the chalcopyrite concentrated in the vein tips) are common throughout the Nova orebody but less abundant throughout Bollinger (Barnes et al. 2020b). The Lower Intrusion thickens to the west from 100 m above the Bollinger orebody to about 200 m thick above the Nova orebody (Fig. 2b).

The Upper Intrusion is a bowl-shaped layered intrusion made up of a large inter-layered pile of meso- to adcumulate lherzolite and olivine gabbronorite, with a lower marginal zone of more orthocumulate rocks.

The intrusions were interpreted by Taranovic et al. (2022a, b) to have intruded initially as a bifurcated sill within a conduit system; sulfide liquid was physically trapped within the lower limb, which became the Lower Intrusion, while magma injection continued into the upper limb over a more extended period, causing it to inflate and develop into a larger layered intrusion. The intrusions were likely emplaced into high-temperature country rocks during the regional deformation and peak granulite-facies metamorphism, as indicated by the sulfide textures (Barnes et al. 2020b), pyroxene chemistry, and magmatic-metamorphic coronas in plagioclase-olivine assemblages (Torres-Rodriguez et al. 2021).

Pyroxenes, both orthopyroxene (enstatite) and clinopyroxene (diopside to augite), are extremely common throughout both the Upper and Lower Intrusions. The pyroxene forms prominent cumulus phases along with olivine, with some units containing very coarse grained (> 1 cm) orthopyroxene oikocrysts.

As a result of the emplacement history outlined above, the Nova-Bollinger orebodies are located below the larger, essentially barren Upper Intrusion (Fig. 2b). This offers an additional challenge to exploration; drilling to the apparent base of the Upper Intrusion in the east (Fig. 2b) would miss the sulfide accumulation in the connected Lower Intrusion. In this study we will investigate the extent to which the complexly zoned pyroxenes extend through the entire intrusive system and make inferences on how effective this tool may be in active exploration.

Metamorphism and alteration

The Upper and Lower Intrusions at Nova-Bollinger were emplaced under granulite facies conditions. Textures and mineralogy, specifically the presence of reaction symplectites, show the imprint of a prolonged period of equilibration under conditions at around 800–850˚C, under predominantly very dry conditions (Torres-Rodriguez et al. 2021). Pyroxenes enriched in Cr typically show evidence of fine-scale lamellar exsolution of Cr spinel, probably developed under near-solidus conditions, with very minor retrograde hydration to amphibole along cleavages.

Sample collection and processing

Polished thin sections and 25 mm diameter polished rounds were taken from diamond drill core samples from pyroxene-bearing lithologies including lherzolite and websterite/pyroxenite from the Upper and Lower intrusions and some gabbronorite and norite samples from the Upper Intrusion. All of these samples have been described in previous studies (Barnes et al. 2020b; Taranovic et al. 2022b; Torres-Rodriguez et al. 2021).

The XRF analysis was focused on the drill holes that make up the E-W cross section outlined in Fig. 2 to ensure a good understanding of the spatial context of the samples in regard to the vertical distance from the main orebody. This includes drillholes SFRD0058 and SFRD0203 which intersect the main massive sulfide unit of the Nova orebody, as well as the upper and Lower Intrusions (Fig. 2).

Kevitsa

The Kevitsa intrusion is located in Lapland, Northern Finland and hosts a Ni-Cu-PGE-Au sulfide ore body. The Kevitsa deposit is described in detail in Santaguida et al. (2015) and discussed in a number of other publications (Le Vaillant et al. 2016, 2017; Luolavirta et al. 2018a; Yang et al. 2013) and is briefly summarized below.

The Kevitsa intrusion is located within the Central Lapland Greenstone belt together with a suite of small- to medium-sized mafic–ultramafic intrusions, ranging in age from ~ 2.44 to ~ 2.05 Ga (Hanski and Huhma 2005). The 2.05 Ga mafic–ultramafic phase includes the Kevitsa intrusion and also the neighbouring Sakatti intrusion, hosting the world-class Cu-Ni Sakatti deposit. The Kevitsa intrusion was emplaced into a volcanic-sedimentary suite comprising komatiites, picrites, and metasediments including sulfidic black shales. The intrusion occupies a surface area of approximately 16 km2 and consists of a lower ultramafic unit up to 1.5 km in thickness, overlain by a 500 m thick mafic unit. The ultramafic unit is composed of interlayered olivine pyroxenite and pyroxenite, with local development of cyclic units, but for the most part lacking obvious internal layering (Santaguida et al. 2015; Le Vaillant et al. 2017). The Kevitsa Ni-Cu-PGE deposit occurs in the middle part of the ultramafic unit (Fig. 3), hosted primarily by olivine pyroxenites. Most of the samples included in this study are clinopyroxene-olivine mesocumulates with variable modes of olivine and augite and variable proportions of orthopyroxene (enstatite) as oikocrysts or transitional cumulus to poikilitic grains.

Fig. 3
figure 3

a Map of the Kevitsa intrusion, with the open pit outlined. Data modified from Geological Survey of Finland (GTK) (2020), created using QGIS. Yellow dots are drill hole collars. b N-S cross section (looking west) through the open pit, with analysed samples marked. Schematic ore-type shells from Le Vaillant et al. (2017) defined from whole rock metal content, where Normal Ore (Ni tenor 4–7%) (yellow and green) with stratification of the PGE tenors, where PGEs are more enriched higher in the stratigraphy (yellow). Ni-PGE Ore (red) is characterised by a Ni/Cu ration of between 1.5–15 and Ni tenors in excess of 30%. False ore (blue) is sulfide bearing zone with very low Ni tenor. Grey outline is extend of Kevitsa intrusion and white is surrounding paraschist. Pyroxene zoning types are described in detail in the results

Full size image

The deposit consists entirely of disseminated sulfides with widely varying Ni, Cu, and PGE tenors. It has been mined since 2012 and the published resources were 237 Mt at 0.25% Ni, 0.34% Cu and 0.41 g/t PGE (Pt + Pd + Au), and 0.01% Co in 2019 (SRK Consulting 2019). The mineralisation has been separated into different ore types, mainly on the basis of their Ni tenors (Mutanen 1997; Santaguida et al. 2015; Le Vaillant et al. 2017). “Normal Ore” represents the bulk (> 90%) of the economic resource and is characterised by Ni-tenor of 4–7%, 2–6 vol.% of sulfides (pyrrhotite, pentlandite, and chalcopyrite), and average Ni and Cu ore-grades of 0.3 and 0.4 wt.%, respectively (Santaguida et al. 2015). PGE tenor is variable, yet on a large scale, higher tenors tend to occur higher in the stratigraphy (Le Vaillant et al. 2017) (Fig. 3b). The volumetrically minor ‘‘Ni–PGE Ore” has a similar sulfide content to that of the Normal Ore, but the sulfides are predominantly pentlandite, pyrite, and millerite, and the ores have higher and more variable Ni grades, lower Cu grades (Ni/Cu = 1.5–15), and extreme Ni tenors in excess of 30% (Barnes et al. 2013). Pockets of Ni-PGE Ore are developed throughout the orebody but the main development is in the upper parts (Le Vaillant et al. 2017). In addition, there are uneconomic, pyrrhotite-dominated low-tenor mineralisation types called “False Ore”. The False Ores occur associated with, or as independent bodies around the main orebodies.

Metamorphism and alteration

The Kevitsa intrusion and orebody have undergone localised retrograde alteration by hydrothermal fluids associated with late-stage normal faults (Le Vaillant et al. 2016). The alteration did not change the overall nickel tenor of the ore, with whole-rock metasomatic effects being restricted to mobile elements Cu and K (Le Vaillant et al. 2016). However, it produced highly variable but locally complete hydration of pyroxene to amphibole: predominantly tremolite-actinolite in clinopyroxene, and cummingtonite-grunerite in orthopyroxene (ESM 1 – Table 2). This replacement is primarily epitaxial, originating along the pyroxene cleavage and progressing to overtake the entire grain in extreme cases.

As the retrogression of igneous pyroxene to hydrothermal amphibole is common in altered mafic–ultramafic intrusions, it is important to understand how this hydration may affect any complex zoning of pyroxene to determine if such zoning has potential as a prospectivity indicator. If zoning survives the hydration process, this would broaden the applicability of this methodology in exploration and research.

Sample collection and processing

27 thin sections were scanned on the XFM beamline at the Australian Synchrotron. These samples represent False Ore (5), Ni-PGE Ore (7), Normal Ore (10); classified based on whole rock assay and 5 samples that do not contain ore levels of sulfides. The location of many of the samples is outlined in the cross section in Fig. 3b.

Three samples, one representing each ore type (Normal, Ni-PGE, and False), were chosen for detailed investigation using the EPMA. Full tables of the major element chemistry of the minerals analysed in this study are included as electronic supplementary material (ESM1 – Table 1). Major element compositions of the pyroxene, olivine, amphibole, and plagioclase are supplemented with additional EPMA data for pyroxene and olivine published in Yang et al. (2013) and Luolavirta et al. (2018b).

Additional samples from the till above the Kevitsa mine (before the open pit was dug) and surrounding area (after the open pit was dug), sieved for the > 45 µm and < 1 mm fraction and separated for heavy minerals using heavy liquid (tetrabromoethane (TBE)) separation to extract all minerals with a specific gravity > 2.95. The heavy mineral separates from the Kevitsa till were dominated by spinel (chromite/magnetite) and pyroxene phases that appear very fresh with minimal alteration or weathering (Fig. 4).

Fig. 4
figure 4

Till sample from above the Kevitsa deposit, separated with heavy liquid techniques. Many pyroxene grains were found in these till and four grains showed abrupt zonation patterns. a Mineral Map scanned using TIMA automated mineralogy (SEM). b XRF false colour image of clinopyroxene with red = Cr, green = Fe, blue = Ca

Full size image

Analytical methods

XRF mappers

This study uses various forms of X-ray Fluorescence (XRF) mapping, including desktop XRF mapping, Maia Mapping, and Synchrotron X-ray fluorescence microscopy (XFM) mapping. All the techniques work on the same principle but differ in beamsize, step size, total map speed, and lower limit sensitivity. The smaller step size (i.e. pixel size) increases the collection time significantly. Details of each XRF method is included within ESM2 – detailed analytical methods.

EPMA

Electron probe microanalysis (EPMA) was carried out using a JEOL JXA-8500F Field-Emission EPMA at the University of Western Australia. For pyroxene and amphibole, an accelerating voltage of 20 keV was used, with a beam current of 20nA and a focused beam, whereas the plagioclase was measured with an accelerating voltage of 15 keV, 15nA beam current, and a 10 µm diameter defocussed spot. Elements were measured using WDS. Oxygen was calculated by stoichiometry.

EBSD

Electron backscattered diffraction (EBSD) analysis can determine the crystallographic orientation of crystals in samples. EBSD analyses were performed on a Zeiss Ultra Plus Field Emission SEM which is equipped with an Oxford Instruments Symmetry EBSD detector at CSIRO Mineral Resources (Perth, WA). Details on the analytical conditions and sample preparation are included in ESM2 – detailed analytical methods.

Results

Mineral compositions

Nova-Bollinger

Clinopyroxene from Nova-Bollinger is diopside to augite in composition, while the orthopyroxene is enstatite (Supp. Figure 13). Both pyroxenes record high Al2O3 levels (clinopyroxene 1.82– 6.03 wt%; orthopyroxene 1.02–3.8 wt%), reflective of their high-pressure crystallisation (Taranovic et al. 2022a, b). Both show a positive correlation between the alumina and magnesium number, which may indicate a continual crystallisation of a single magma source. The few outliers of very high Al content may be due to the measurement of aluminous spinel exsolutions. The Ni, Ti, and Cr contents show no correlation with the magnesium number (Supp. Figure 13).

Kevitsa

The pyroxene assemblage throughout the Kevitsa intrusions is dominated by clinopyroxene, with rare oikocrystic orthopyroxene. The Cr content of the pyroxene ranges from 8,000 ppm in the enriched zones of the clinopyroxene to lower than the estimated limit of detection (EPMA 200 ppm) in the Cr-poor zones (Supp. Figure 14). The orthopyroxene has maximum Cr values of 4,000 ppm.

The clinopyroxene is diopside to augite in composition (Supp. Figure 14). Clinopyroxene in the Ni-PGE Ore is almost exclusively diopside and has chemistry that is more Ca- and Mg-rich than the False and Normal Ore types. The orthopyroxene is enstatite in composition. Clinopyroxene is also abundant in the till above the Kevitsa intrusion (Fig. 4). The major element chemistry of the ex-situ pyroxene mimics those within the intrusion.

As has been demonstrated in previous studies, the olivine within the Ni-PGE Ore is extremely Ni-rich; up to 14,000 ppm Ni (i.e. 1.4 wt% Ni) (Luolavirta et al. 2018b; Yang et al. 2013) and has a large range with an average 8,871 ppm ± 3,389 ppm (n = 179), showing a negative correlation with Fo content (Barnes et al. 2023). Olivine in False Ore samples has very low values of Ni, with an average of 301 ppm ± 246 ppm (n = 50) (Luolavirta et al. 2018b). The olivine in the Normal Ore has moderate values of Ni with an average of 1,504 ppm ± 615 ppm (n = 334) (Luolavirta et al. 2018b). Similarly, the clinopyroxene and orthopyroxene in the Ni-PGE Ore of Kevitsa also shows high Ni content (average 739 ± 482 ppm and 1,000 ± 697 ppm respectively) compared to the other ore types (< 500 ppm) (Supp. Figure 15).

The amphibole compositions were measured in this study and recorded along with their protolith mineral. The amphibole nomenclature is derived using the Locock (2014) method (refer to ESM Table 2). The clinopyroxene from the Ni-PGE Ore converted to a magnesio-hastingsite composition. Both pyroxenes within the Normal Ore sample transform to an edenite amphibole, while some amphibole grains were interstitial to the large pyroxene and are more magnesio-hastingsite in composition.

Pyroxene zoning throughout the Nova-Bollinger intrusions

Drillhole SFRD0058 represents the thickest section of the Lower Intrusion, whereas SFRD0203 represents a typical section through the centre of the deposit (Fig. 5). Cr zoning is observed through the entire intrusion and summarized symbolically in Fig. 2.

Fig. 5
figure 5

Detailed core logs with sample locations (yellow diamonds) and XRF images of zoned pyroxene displayed as R.G.B = Cr.Fe.Ca false colour images where clinopyroxene is shown as pink/purple and orthopyroxene as orange. Olivine is bright green. a-c drillhole SFRD0058 through the Upper Intrusion, the surrounding gneiss and into the Lower Intrusion that hosts the Nova orebody. Massive sulfide is found at the base of this drillhole with net-textured sulfide extending from 295 m to the massive sulfide at 343 m. Disseminated sulfides are present from 271-295 m. Samples were taken in the upper regions of the Lower Intrusion, near the contact with the surrounding gneiss at 187 m and 211 m. Thin pegmatites are also present in this drillhole (pink). d-f XRF images of samples from the Lower Intrusion in hole SFRD0203. Both orthopyroxene and clinopyroxene show sector zoning (e) and abrupt zoning (d,f). g SFRD0203 drilllog which intersects the Upper Intrusions, the surrounding gneiss and the Lower Intrusions and the Nova orebody. The massive sulfide has separated from the Lower Intrusion and intruded into the garnet gneiss. h,j Samples from the Upper Intrusion, XRF images showing no zonation in the clino- and orthopyroxene and i some normal zoning in the clinopyroxene

Full size image

Samples from the massive sulfide layer show some abrupt Cr-zoning in both ortho- and clinopyroxene (Fig. 5F). Samples, such as SFRD0058-211 m in the thickest section of the Lower Intrusion (Fig. 5a), show Cr-rich cores of both clinopyroxene and orthopyroxene and both have abrupt decrease in Cr content between the core and rim. The clinopyroxene appears normally zoned (i.e. diffuse decrease in Cr content toward the rims) when measured with the lower resolution XRF mapper but appears much more abrupt using the very high resolution XFM beamline at the Australian Synchrotron. This section contains no visible disseminated sulfides and is approximately 150 vertical meters away from the massive sulfide intercept and 85 m away from any disseminated sulfide.

The orthopyroxene zoning patterns at the base of the Lower Intrusion show both sector zoning with abrupt changes in the core to the rim (Fig. 5d, e). Sample SFRD0203-343 m is from the middle of the Lower Intrusion (Fig. 5d, g) and displays sector zoning and rarely abrupt zoning. Clinopyroxene is less common and shows Cr-rich cores that transition to Cr-poor rims but it is difficult to determine if this is ‘abrupt’ or continuous zoning related to fractional crystallisation of the trapped liquid (Fig. 5a).

The sector zoning in many of the orthopyroxene grains in the Lower Intrusion can be visualised in thin section (Fig. 6) due to the fine scale spinel exsolution. The sector zones with higher Cr content contain denser spinel exsolutions giving a darker colour when viewed under plane polarised light.

Fig. 6
figure 6

a schematic diagram of an orthopyroxene crystal with sector zoning and spinel exsolution along the favoured crystal axis. b XRF image of orthopyroxene in from sample SFRD0203-343 m. False colour image where red = Cr, blue = Ca and green = Fe. c Same region as (b) in plane polarised thin section imagery. Sector zoning is visible due to increased abundance of spinel exsolution in the Cr-rich sector zones. d EBSD-inverse pole figures from a small region of this grain show the x, y, z axis of the sample are aligned with < 001 > , < 100 >  < 010 > direction of the crystal. This further confirms that the exsolution is parallel with the (001) and (010) planes

Full size image

Samples from the base of the Upper Intrusion and the C5 intrusion show some sector zoning and abrupt zoning in the orthopyroxene, whereas the clinopyroxene does not show any distinct zoning pattern (Fig. 2).

The Upper Intrusion is made up of interlayered gabbronorite/norite and lherzolite/feldspathic lherzolite. The pyroxene in the upper units is commonly unzoned with only minor normal zonation even when examined using fine scale and low detection limits of the XFM beamline. The pyroxene in the lherzolite at the very top of the intrusion (Fig. 5h, i, j) show no zoning at all.

Symplectites are common throughout the Upper and Lower intrusions (Torres-Rodriguez et al. 2021) are formed due to a decompression reaction between neighbouring plagioclase and olivine crystals. This reaction creates orthopyroxene as a bioproduct which can also display normal zonation (Supp. Figure 16).

Cr-Ti zoning types in Kevitsa

The Cr zoning patterns in the clinopyroxene are extremely complex (Fig. 7). Cr and Ti zoning is ubiquitous in the entire sample suite and can be characterised in terms of three main types (1) “crater”, (2) “moat”, and (3) oscillatory, which are defined by variation in Cr content and can occur in combination with one another (Fig. 7).

Fig. 7
figure 7

New pyroxene zoning types found in Kevitsa. (a-c) crater zoning and (d-f) Moat zoning. a XRF image of clinopyroxene grain in sample KV280-547.95 m (False Ore type) which shows crater zonation, chemical traverse is measured along the white arrow. XRF false colour image is red = Cr, green = Mn and blue = Ti. b changes in Cr and Ti content along the traverse shown in A. Note that increases in Cr correlated to a decrease in Ti and vice versa. c Schematic diagram of crater zoning. d XRF image of a clinopyroxene grain in sample KV185-133.21 m (Normal Ore type) that shows moat + crater zoning. Chemical traverse is measured along the white arrow. XRF false colour image is red = Cr, green = Mn and blue = Ti. e changes in Cr and Ti content along the traverse outlined in D. Note that the decrease in Cr is not paired with a significant aberration of Ti content. f schematic diagram of moat zoning

Full size image

The most common zoning style is termed “crater”, based on the presence of a homogeneous Cr-poor core, surrounded by a sharply bounded concentric Cr-rich zone, dropping off to a Cr-poor margin (Fig. 7a, b, c). The peak Cr content also commonly correlates with the lowest Ti values. These crater-zoned crystals can also display a “moat” type zoning subtype (Fig. 7d). The moat subtype refers to a significant drop in the Cr content, bounded on both sides by very high Cr values (Fig. 7d, e, f).

The outermost zone typically has decreasing Cr and increasing Ti contents, zoning out into a homogeneous Cr-poor and Ti-enriched outer rim. The outer rims of the clinopyroxene commonly display a number of concentric zones of higher and lower Cr (oscillatory zoning). In the common case where the plane of the thin section is not sliced through the relatively small pyroxene core, grains appear to show normal (high-Cr core to low-Cr rim) (Fig. 1-B1) and/or oscillatory zoning (Fig. 1d).

At very fine scale (2 μm resolution) the Cr-spinel exsolution in the Cr-rich zones is observed following the crystal planes, highlighted by the twin planes observed in Fig. 8c. This fine exsolution lamellae of a Cr-enriched spinel is usually developed along the (100) lattice plane (Savelieva et al. 2016) as confirmed by EBSD analysis (Fig. 8f). This exsolution is particularly well developed within the Cr-enriched portions of the zoned clinopyroxene grains (Fig. 8f). The Cr-spinel exsolution is easily observed in thin section (Fig. 8, Supp. Figure 17).

Fig. 8
figure 8

Exsolution along the crystal planes of a clinopyroxene in sample KV185-133.21 m. a XRF image of the clinopyroxene grain XRF false colour image is red = Cr, green = Mn and blue = Ti. b XRF image of Ca content, OPX exsolution is signified by a near zero calcium content (black). c XRF image of the clinopyroxene grain XRF false colour image is red = Cr, green = Mn and blue = Ca. Change in angle of the spinel exsolution signifies a twin plane in the crystal. d a high resolution backscattered electron image (BSEI) showing this change in the alignment of the spinel. Only spinel that intersect the surface are shown. e the spinel exsolution as visible in plane polarised light in thin section. Spinel under the surface are visible using this technique. f EBSD-inverse pole figures of a small area of this grain. The grain is visualised on the (010) plane with the (-100) near parallel with the x axis of the image and (001) near parallel with the Y axis. g a schematic diagram of a clinopyroxene grain with the orthopyroxene and spinel exsolution shown

Full size image

The orthopyroxene forms large euhedral oikocrysts containing inclusions of olivine and clinopyroxene. Typically, the poikilitic or transitional poikilitic-cumulus orthopyroxene is not intensely zoned, but shows subtle normal zoning (high-Cr to low, low-Ti to high) from core to rim and sector zoning (Supp. Figure 18).

Ni-PGE Ore

The clinopyroxene grains in the Ni-PGE Ore show abrupt zoning with Cr-rich cores surrounded by low Cr, Ti rims (Fig. 9a, b). These are overgrown by late stage, interstitial clinopyroxene with moderate Cr values. Notably, crater-type clinopyroxene grains occur less frequently than in the other types. The outer edges of grains show Ni enrichment, correlating with the reduction in Cr contents (Supp. Figure 19). Within the Ni-PGE Ore, orthopyroxene is rare (Mutanen 1997; Luolavirta et al. 2018b) and when present displays no zoning. Most of the samples are significantly altered, with parts of the thin section showing patchy epitaxial amphibole alteration (Fig. 9a) where the original Cr zonation is still visible through a combination of the spinel exsolution resisting the amphibole alteration and the amphibole composition reflecting the original pyroxene composition. This zoning is known as “ghost zoning” and is found commonly in the Ni-PGE ore but also occurs in any sample with weak to moderate amphibole alteration (Fig. 10).

Fig. 9
figure 9

Summary of the zoning in the clinopyroxene of Kevitsa from the various ore types. All images are XRF false colour images with the colours red = Cr, green = Mn and blue = Ti. Field of view is 5 mm2 in all images. a Ni-PGE Ore, sample KVNMEI-82.55 m with ghost zoning (previous crater zoned) in clinopyroxene that is strongly altered to amphibole. b Ni-PGE Ore, sample R695-67.8 m with strong Cr-rich cores and abrupt zoning with moderate amphibole alteration. c Normal Ore (with low PGE) from sample R807-621.05 m with pristine clinopyroxene with crater zoning. d Normal Ore (with high PGE) from sample KV185-133.21 m with moderate amphibole alteration and clinopyroxene grains with crater zoning and a distinct Ti enrichment in the rims. e False Ore from samples KV280-620.4 m with crater zoning in the clinopyroxene and no zoning within the orthopyroxene. This sample has very little amphibole alteration. f False Ore from sample KV280-547.95 m with pristine clinopyroxene grains with some strong crater zoning and some moat zoning

Full size image
Fig. 10
figure 10

Progressive alteration of clinopyroxene to amphibole. XRF false colour images where Cr = red, Mn = green and Ti = blue. a Sample KV103-687.74 m (Ni-PGE transitional ore) with clinopyroxene (cpx) that shows relic crater zoning, now ghost zoning, with the amphibole (amp) inheriting the Cr-rich and Cr-poor sections from the original clinopyroxene. b a clinopyroxene grain from sample R695-67.80 m (Ni-PGE ore) with abrupt zonation between Cr-rich core and poor rims with minor amphibole alteration. Amphibole is patchy throughout the grain. c A number of clinopyroxene grains from sample KVNMEI-82.55 m (Ni-PGE ore) with ghost zoning of relic crater zones. Amphibole alteration is moderate in this sample. d KVNMEI-88.92 m (Ni-PGE ore) sample of clinopyroxene with ghost zones of abrupt zoned clinopyroxene. The amphibole alteration in this sample is moderate. e An XRF map of a thin section from sample KV309-142.8 m (unmineralised) with fresh clinopyroxene (right) and heavily altered clinopyroxene/amphibole (left). Some ghost zonation is visible however as the alteration progresses this zoning is overprinted. Orthopyroxene (opx) remains relatively unaltered in most samples, even with pervasive alteration of clinopyroxene to amphibole

Full size image

Normal ore

The Normal Ore samples examined in this study consist of dominantly clinopyroxene with some olivine, and less abundant orthopyroxene (Fig. 9c, d). Clinopyroxene displays crater zoning, with Cr-poor cores surrounded by large Cr-rich zones. Frequently, clinopyroxenes are sectioned through the outer zones of the crystal only, showing an apparent “abrupt” zonation, however, it is likely that these grains contain the crater signature in their cores. Sample KV185-133.21 m also shows very low Cr and high Ti rims around all clinopyroxene grains. Amphibole alteration is weak to moderate in these samples (Fig. 9d).

False ore

The False Ore samples consist of clinopyroxene, olivine and poikilitic orthopyroxene. Clinopyroxenes are dominated by crater zoning (Fig. 9e, f). Both cumulate and interstitial growth of clinopyroxene are common. False Ore sample KVX014-643.4 m (Fig. 11) is a good example of the extremely complex zoning in the clinopyroxene that can be present in a single sample. The large orthopyroxenes are poikilitic, containing olivine and sulfide inclusions. Furthermore, there are reaction rims surrounding the orthopyroxenes that are infilled by plagioclase (Fig. 11b), where the clinopyroxene appears to dissolve at the expense of the orthopyroxene growth. Among the studied thin sections, the False Ore samples are generally less altered than the other ore types. The non ore parts of the intrusion have pyroxene zoning patterns that are indistinguishable from the False Ore zoning patterns.

Fig. 11
figure 11

Synchrotron XRF image of a thin section from sample KVX014-643.4 m, which is part of the olivine (ol) websterite “False Ore” zone. False colour maps where Cr = red, Mn = green, Ti = blue. a Overview scan of the entire thin section. b The large orthopyroxene (opx) oikocrysts are grown at the expense of clinopyroxene (cpx), leaving dissolution haloes around the opx grains that are infilled with plagioclase (pl). c Sector zoning of Cr is visible in the opx. d Extremely complex zoning of Cr in cpx with at least 4 growth stages (labelled 1–4) and highlighted in the chemical traverse, with the amphibole (amp) alteration labelled as stage [5]. e another example of the 4 growth stages of the clinopyroxene and chemical traverse through the grain. Note that the titanium content has a smooth reduction toward the edges of the grain and a large enrichment at the very edges of the grain, while the chromium content exhibits much larger changes throughout the crystal growth

Full size image

Glacial till

Clinopyroxene was also found commonly within the glacial till samples collected from above and surrounding the Kevitsa open pit. The average volume of clinopyroxene measured in a 25 mm diameter epoxy round was 23 vol% (± 15 vol% standard deviation) while orthopyroxene is much less abundant at an average of 2 vol%. The till samples also commonly contain magnetite, ilmenite, chromite, amphibole and olivine. Within the 20 till samples collected, abrupt zonation in clinopyroxene was observed clearly in 1–4 grains in three samples. This zonation was observed using the scanning electron microscope for automated mineralogy (i.e. TIMA—see ESM 2 – detailed analytical methods) which has low counting times. If the samples were analysed using XRF techniques, the zoning would be more easily observed and is likely more ubiquitous than suggested by this technique.

Discussion

Recording magma histories in pyroxene grains

The chemical zoning patterns within the clinopyroxene grains at Kevitsa are extremely complex. New zoning types that include moat and crater zoning have been defined in this study and have not been commonly observed in other intrusion hosted ore deposits. The clinopyroxene grains throughout the False Ore and Normal Ore are dominated by grains with low Cr contents in the cores, cyclic enrichment in the mantles of grains, and commonly ending in a rim of low Cr paired with high Ti values.

Complex, high-amplitude oscillatory zoning of chromium as reported here is unusual in layered intrusions (Schoneveld et al. 2020; Jenkins et al. 2022). It has been observed in volcanic rocks, particularly in andesitic eruption products of arc volcanoes (Ubide et al. 2015, 2019; Ubide and Kamber 2018) and in alkali basalts (Welsch et al. 2016) where it is interpreted as the result of magma mixing in subvolcanic reservoirs. Kevitsa appears to be an example of this process that is preserved in a tholeiitic system, consistent with the interpretation of Luolavirta et al. (2018a, b) of Kevitsa as a multiply replenished chamber.

As an example, a clinopyroxene grain from sample KVX014 (Fig. 11) shows that the first stage is represented by (1) cores of the pyroxene with low Cr (1,000 ppm) and moderate Ti (2,300 ppm). The second stage of growth (2) in the clinopyroxene has moderate Cr (4,200 ppm) that decreases to 3,300 ppm while the Ti values are very low < 100 ppm. The third growth stage (3) shows very high Cr content (6,500 ppm) with no change in Ti (< 100 ppm), and some undulating boundaries suggest resorption of the previous growth stage. The fourth (4) and final stage shows steadily decreasing Cr values (3,900 → 3,100 ppm) and increasing Ti (< 100 → 2,500 ppm). This final stage likely records the final “trapped liquid” reactions of the mostly solid rock, where the compatible elements (Cr) decrease and the incompatible elements (Ti) increase as the cumulus grains react with evolving trapped intercumulus liquid. These zonation patterns are not reflected in the large orthopyroxene oikocrysts. The final stage (5) represents the alteration of clinopyroxene to amphibole.

Many of the samples, particularly from False Ore, show a dissolution reaction of clinopyroxene at the expense of the growing orthopyroxene (Fig. 11b). The cores of these large orthopyroxenes show no evidence of Cr depletion. There is rare moat zoning (i.e.depletion in Cr in the mantle of the grain) but generally, the Cr content of the orthopyroxene is consistent. This disparity between the zoning styles in the pyroxene species suggests different crystallisation histories.

The dissolution of clinopyroxene at the expense of orthopyroxene growth indicates a peritectic reaction. Intriguingly, the exact opposite relationship was observed by Barnes et al. (2021) in a common pyroxenite rock type in the Bushveld Complex, where orthopyroxene was dissolved at the expense of clinopyroxene growth. The texture is otherwise remarkably similar, notably in the olivine-pyroxene depleted “halo” around the oikocrysts, interpreted by Barnes et al. (2021) as the result of dissolution of the resorbing phases (clinopyroxene and olivine in this case) in a grain-scale boundary layer ahead of the growing oikocryst. The lack of oscillatory Cr zoning in the oikocrysts implies that these grew under static conditions in the mush after the zoning in the clinopyroxene had already formed.

The combination of the widespread oscillatory zoning, dissolution textures of the clinopyroxene, the large grain size of the orthopyroxene and the disparity in zoning styles may indicate that the Kevitsa region underwent multiple late-stage injection or percolation of magma through a pre-existing crystal mush. New injections of magma into a thick crystal mush pile may have caused the entire mush pile to re-fluidise and may cause subsequent overturn of the pile and additional mixing (Fig. 12) (Schleicher et al. 2016). This re-injection and mixing may be an explanation of the minimal amount of massive sulfide at the base of Kevitsa, as any accumulated sulfide would be re-entrained by the mixing of the new magma injections and disseminated through the crystal mush. Progressive increase in Ni and PGE tenor through the Normal Ore may reflect upgrading at high magma-sulfide ratios during this process.

Fig. 12
figure 12

Simulation of the injection of new basaltic magma into a crystal mush pile from Schleicher et al. (2016) modified to describe the potential mixing within the Kevitsa intrusion with colour scheme to match Fig. 3b. a A clinopyroxene mush pile (40% porosity) at the base of a magma chamber. Melt within the intrusion is black, new magma is red. b injection of new, Ni-PGE laden magma (red) into the base of the crystal pile causes mixing and induces a viscoplastic response. The mixing between the layered crystal pile and the new magma cause the ore zones and cause of the complex zoning patterns in the Kevitsa clinopyroxene. Areas of high volume of new magma, in the turbulent eddies, may result in the very high Ni tenors throughout Kevitsa

Full size image

A characteristic feature of the Ni-PGE rich ore, in addition to its extremely high Ni tenors in sulfide, is the presence of exceptionally Ni-rich olivines (Yang et al. 2013; Barnes et al. 2023) and pyroxene. Barnes et al. (2023) attribute these to the entrainment into a relatively low Mg/Fe magma of pre-existing Ni-rich sulfide derived originally from a komatiitic magma with much higher Mg/Fe. The equilibrium exchange of Fe and Ni between the entrained sulfide and olivine derived from the low Mg/Fe magma cause the olivine to become anomalously Ni-enriched. This process is rapid owing to the fast solid state diffusion rate of Ni in olivine with diffusion coefficients of up to logD−14.5 for Ni (Spandler and O’Neill 2009), equating to a diffusion profile of ~ 33 µm per day at 1300˚C which is a similar rate to Mg and Fe. Similarly, Ni-rich outer rims are observed in the clinopyroxene (Supp. Figure 19) which could be due to either the solid-state diffusion or additional growth in a Ni-enriched magma.

Extent of pyroxene zoning throughout conduits

Cr zoning in pyroxene is common throughout the Nova-Bollinger and Kevitsa intrusions, particularly within the ore zones themselves, and through many magmatic sulfide deposits globally (Schoneveld et al. 2020). However, the vertical and/or lateral continuation of this prospectivity signature has not been studied in detail. The large volume of unmineralised pyroxene-bearing cumulates in the host intrusions at Nova-Bollinger provide an excellent opportunity to investigate.

The pyroxene within the sulfide-bearing layer and olivine-rich lherzolite and pyroxenite of the Lower Intrusion at Nova-Bollinger show complex zoning in almost every orthopyroxene and clinopyroxene grain. Cr-rich cores characterise pyroxene grains as far as 150 m vertically removed from the massive mineralisation and 85 m above the disseminated sulfide zone, although oscillatory or sector zoning is not recorded.

Pyroxene in the Upper Intrusion at Nova-Bollinger lacks sector or abrupt zoning, although sector zoning is recognised in orthopyroxene within the basal gabbronorite pyroxenite unit (Fig. 2b), and normal zoning of Cr-poor and Ti–rich rims is observed within oikocrysts from poikilitic peridotite layers (Taranovic et al. 2022a, b).

Complex intra-grain zonations within pyroxene are likely the result of a dynamic conduit system that allows high magma through-put and incorporation and melting of surrounding wall rock, which can cause changes in chemistry of the magma. The observations of the pyroxene zoning at Nova suggest that the Lower (ore-bearing) Intrusion may have incorporated wall-rock into the crystalising magma during the emplacement of the sulfides, a process that extends even to the upper parts of the Lower Intrusion.

The Upper Intrusion is a thick sequence (up to 400 m thick) that likely required relatively longer times to cool and crystallise; the studied samples from the Upper Intrusion are from the interior of the body, well away (50-70 m) from wall rocks. Consequently, significant zonation in pyroxene chemistry is only reported within the basal parts of the Upper Intrusion, or the C5 unit that forms the intersection between the Upper and Lower intrusions (with disseminated sulfide). The lower parts of the Upper Intrusion probably developed first, simultaneously with the Lower Intrusion, prior to subsequent inflation to form the main upper chamber (Taranovic et al. 2022b) where cooling rates were much slower, and pyroxenes were able to grow under equilibrium conditions of low supercooling. However, as noted by Barnes et al. (2023), olivines several hundred metres above the base of the Upper Intrusion show Ni depletion signals consistent with prior extraction of sulfide. This is interpreted as the result of continuing supply of the inflating Upper chamber with previously Ni-depleted magma derived from the earlier mineralising event. The depletion of nickel in olivine seems to be a more widespread indicator of the sulfide interaction, whereas the pyroxene probably reflects the transition from an open system in the Lower Intrusion to a closed system in the Upper Intrusion.

At Kevitsa, minor element zonation in pyroxene appears to be almost ubiquitous throughout the intrusion. The best developed and most complex crater and moat zonation patterns are developed in pyroxene within the False Ore and non-ore zones; these pyroxenes are also generally the least altered, with only weak alteration to amphibole. A possible explanation is that the Kevitsa intrusion developed largely as a crystal-rich mush that was regularly disturbed and stirred up by influxes of new magma through a fluidised mush pile, similar to the process modelled by Schleicher et al. (2016) (Fig. 12).

Robustness of Cr zoning in pyroxene through alteration to amphibole

The apparent patchy nature of the amphibole alteration observed within the Kevitsa deposit may be due to the orientation of the pyroxene. The amphibole exsolution begins the replacement as plates along a cleavage related to a preferred crystal plane, generally the (010) plane in clinopyroxene and the (100) plane in orthopyroxene (Chashchukhin and Votyakov 2012; Smith 1977). Similarly, the zonation within the pyroxene is crystal orientation dependant. Sector zoning specifically is visible on the (010) face due to the differing partition coefficients on the perpendicular faces. The spinel exsolution is also controlled by crystal axis, generally (100) in diopside and (010) in enstatite (Savelieva et al. 2016), which was confirmed in this study by EBSD.

Although the spinel lamellae are thought to be controlled by the (100) axis, they take the form of long rods that are also aligned with the (010) axis. This may be why the amphibole alteration is more pervasive in the clinopyroxene over the orthopyroxene, as the exsolved spinel may allow easy access to the preferred alteration axis. The orthopyroxene, however, shows spinel and amphibole with preferred axes that are mutually perpendicular.

When the replacement is partial, and some pyroxene remains, the zonation patterns remain visible through “ghost zoning”, where the amphibole inherits the enriched and depleted trace element patterns from the original clinopyroxene (Fig. 10A-D). As the amphibole alteration becomes more pervasive, the original chemical zonation is wiped out, due to recrystallization of the amphibole and the destruction or conglomeration of the spinel exsolution lamellae that developed within the original Cr-rich zones (Fig. 8).

Pyroxene as an ex-situ indicator

Pyroxene is extremely common in the surface till samples from above and around the Kevitsa deposit with a few of the clinopyroxene grains exhibiting abrupt changes in the chromium contents (Fig. 4). Although this is based on limited study, it suggests that complexly zoned pyroxene may not only be useful as an in-situ indicator, from diamond drilling of the target intrusion, but may extend to an ex-situ indicator in surface samples in regions with glacial till.

Conclusions

Intra-grain chromium zoning in pyroxene is pervasive within the mineralised Lower Intrusion of the Nova-Bollinger deposit (Western Australia), interpreted to indicate that these pyroxenes are derived from high-flux regions of the upstream conduit system. The abrupt-style zonation pattern in Nova-Bollinger is observable in lherzolite up to 150 m vertically from the massive sulfide and 85 m above the disseminated sulfide zone. Zonation is not observed in the unmineralised Upper Intrusion. Therefore, intra-grain chromium zonation in pyroxene is a good indicator for prospectivity as it may vector toward the dynamic parts of intrusive complexes that represent enhanced prospectivity for metal-rich orthomagmatic sulfide ores.

Chromium-zoning in pyroxene is pervasive through the Kevitsa intrusion (northern Finland), likely due to dynamic recharge processes and a complex thermal history of remobilised cumulates. Within the Kevitsa intrusion, Cr-zonation is progressively overprinted and wiped out by hydration of pyroxene to form amphibole. This overprint is progressive such that “ghost zoning” can still be detected at moderate degrees of alteration, but zoning is obliterated when replacement is complete. Therefore, this indicator is robust to moderate alteration and hydration.

Complexly zoned pyroxene has been found in glacial till above the Kevitsa deposits. This suggests that not only is this tool useful as an in-situ indicator mineral but may survive as an ex-situ indicator for more regional exploration.

Observation of complex zoning (i.e. abrupt, sector, crater, moat or oscillatory zoning or any combination thereof) in either clinopyroxene or orthopyroxene implies an active or open system. However, as this is an indicator of magma through-put or cycling, which relates to favourability for upgrading the metal tenor of the ore, rather than a direct indicator of sulfide saturation or accumulation, the absence of complex zoning in the intrusion does not guarantee an absence of sulfides. Therefore, this tool should be used in conjunction with other prospectivity tools as part of a weighted evidence approach.