Regolith Hosted Rare Earth Elements in (?allanite)-apatite-magnetite monzogranites, Koorda area, Western Australia: Mineral System Approach

Regolith-hosted Rare Earth Element deposits (REE's) have, of late, formed an important exploration target for a supply of non-refractory REE's which are not associated with carbonatites (strictly speaking) or ore deposits associated with carbothermal systems and alteration by the classification of (Mitchell & Gittins, 2022).

Typical examples of regolith hosted REE's of economic significance are examples from China, rich in heavy REE's (HREE); deposits in Brazil, and deposits in Madagascar.

Regolith-hosted REE's span a gamut of lithologies and genetic provenance, from detrital accumulations of refractory minerals in heavy mineral sands in palaeoplacer beach deposits or fluvial/alluvial placers, to weathered lithocaps over alkaline ultramafic deposits, to detrital REE-enriched tuffaceous deposits in shallow marine deposits.

The main precondition for a deposit of REE's in the regolith to be economic is not so much grade (though of course extraordinary grade can hide many sins of metallurgy), but it is the nature of REE deportment within the minerals that form the bulk of the rock, and the ionic or colloidal attachment of those REE's to those minerals.

Ideal regolith hosted REE deposits, exemplified by Chinese examples, would be those which are close to surface, with minimal overburden, wherein rare earth elements are bound by weak ionic forces to minerals with high cationic exchange capacity or CEC (typically here one cites kaolinite), and which are susceptible to desolvation from the ionically bound state via a lixivant such as ammonium sulphate, at near-neutral pH. These would be whatever the Chinese word for unicorn is, as there is much made of the ionic clays, and less made of colloidally-bound REEs (and again, grade hides many sins here).

Colloidally bound REE, wherein REE is bound up within (hydr)oxide minerals formed within the regolith, are of lesser perceived value, due to the need to undertake greater chemical work to desolvate or dissolve the hydroxide minerals and their colloidally contained REE. Or, in another way of looking at it, REE which are substituted for 3+ valence elements within a hydroxide mineral, one needs to destroy the mineral, typically with acid.

The REE Rush

In 2018-19 or thereabouts, attention of investors and geologists in Australia set their mind to REE exploration again, as market conditions were favorable, and speculative appetite for 'Critical Minerals', or more precisely, government subsidies and funding for exploration for Critical Minerals, was on the increase. Typically, the same carbonatite deposits that had been explored since the 1960-70's were trotted out and zhuszhed up to enter another round of drilling and resource definition work. But minds were also open to alternative sources such as ionic clay hosted deposits.

Attention piqued, data mining began, and readily highlighted the presence which had been explored for nearly a decade by Salazar Resources Pty Ltd, who had defined a resource of REE bearing clay.

Understanding of the deposit was limited at the time, but nearology being what it was, further investigatory data mining of WAMEX reports from the general region highlighted province-scale TREE anomalies in the Munglinup belt, culminating in SOC Resources Pty Ltd pegging, and being granted, two exploration licenses at Cowalinya, on my recommendation.

Drilling of Cowalinya by eMetals Limited in 2019-20 successfully demonstrated the presence of enriched TREO in lower saprolite clay to saprock clay, over a >8km long traverse of air core drilling along an existing track.

Assay methodology which was chosen at the time was to sample on 2m composite basis, with two subsamples taken, and subject one subsample to a 4-acid digestion and the other to a chloride-based partial leach digestion in 0.1M HCl. The delta between the two, if large, would demonstrate a refractory mineralogy hosted the REE's, and if the delta between the two methods was low, it would demonstrate a possibility for ionic or, at least easily soluble, REE hosted within the regolith.

Results showed TL7 digests recovered approximately 90-95% of the tenor of REE's compared to the 'near-total' 4-acid digest results, which supported the hypothesis that the REE in the regolith was likely 'soluble' and therefore supported the continuance of the project. It also supported the regional and province scale opportunity before eMetals at the time.

Unfortunately for eMetals, the REE enriched clays at Cowalinya were at too great a depth to be immediately attractive, and I moved on from my role consulting to them, with the proof of concept completed. Within months of this result, the entire Munglinup was taken up with tenure, and others benefited, till recently, from the speculative and nearological benefits made from this limited, targeted and parsimonious testing of an immature belt-scale opportunity.

Toward Apatite-Magnetite Granites

In late 2021, I considered that, given the success at data mining in the Munglinup, there was surely another terrane-scale REE opportunity to be unearthed in Western Australia, and set my mind (in my limited spare time) to unearthing further 'ionic' REE terranes in Australia.

Soon, I realised that I would have to do a lot more work than just downloading WAMEX and summing REE's and converting to max downhole TREO+Y or EOH assays. Not that I didn't do this. I did.

I considered that granites were the next-best source of REE's in Western Australia for several reasons;

1. Granites tend to be large (ignoring syenite dykes, which we'll get to later) and can provide volumes of clay sufficient to run large open pit mines with low margins
2. Granites can occasionally become enriched in REE's
3. Granites weather, typically, into kaolin, a mineral good CEC and hence, potentially liable to form a mostly ionically-bound REE comportment
4. There is good data on granites in the GSWA
5. The Yilgarn Craton is mostly granite, most is un-pegged, and considered unprospective.

I was, in 2021-2022, wrong on point 4, despite the ongoing efforts of Hugh Smithies et al. (2020), at the GSWA to increase the data coverage, sampling and geochemical characterisation of granites in the Yilgarn Craton. Here, I remind those readers who have persisted this far, that we now know what was not known 2 to 3 years ago, when the first update of the GSWA Granite Database came out; back immediately before this one had WACHEM and had to sort WACHEM data and perform some considerable mental gymnastics to gain hints that granites could, in their fresh state, breach the magic 750ppm TREO threshold that was in 2021 considered 'good' grades for REE's. I know it's hard, today, to get out of bed for 750ppm TREO, but in 2021-22, that was my threshold.

Armed with the freshest GSWA data in 2022, I resolved to undertake their work for them. First, I needed to understand where in Western Australia highly REE-enriched granites were.

First barrier to entry was incomplete assaying of all REE elements in rocks and granites in the database. Secondly was the understanding that one does not want monazite hosted REE's, as this will skew towards LREE enrichment, and complicate metallurgical performance with unwanted La and excessive Ce; Nd-Pr are fine to have of course, but they will go up in even HREE enriched (c.f., A-type) granitoids, but one would prefer HREE enriched granitoids so as to have a HREE enriched clay and minimal La-Ce penalties.

My first effort, therefore, was to consider ytterbium as a good proxy for TREE enrichment, as even a small incremental increase from 2 to 4 ppm may indicate, in a very thumb-suck way, that the remainder of the REE's would be enriched. In the map below the 2021-22 vintage data highlights the west Yilgarn as the most Yb enriched region; the Eastern Goldfields superterrane has isolated but rich Yb, and these are typically syenites, and they are typically small dykes Please now refer to Kairos Minerals et al., who have subsequently defined moderately enriched syenites; these are also known from Wallaby, and are likely associated with sanukitoid/syenite related gold systems; these were of no interest as regolith REE's to my mind, as they were all under tenure.

Within the AOI, above, I then synthesized a magnetic and gravity interpretation of large granitic plutons. Boundaries are defined by major magnetic and gravity worm boundaries. Interestingly, there appeared to be a fundamental boundary running broadly from Southern Cross-Mukinbudin, to Wubin-Morawa, with a southern margin passing south of Koorda. This will be elaborated upon later.

Gravity worms showed, firstly, that the supracrustal belts which are part of the Corrigin Tectonic Zone, are truncated on the north by a fundamental deep crustal structure, represented by the gravity worms. North of this lies a huge area dominated by weakly magnetic, but deeply low gravity annular and circular granite plutons.

This allowed the following tectonic interpretation to be constructed. This is an interesting food for thought; how do the southern Murchison and the Corrigin Tectonic Zone interact and how was this part of the Yilgarn Craton assembled? Gravity and granitic character data will be presented to show an hitherto unrocognised magmatic and tecotnic association hidden under wheat, salt lakes and ferocious emus.

The initial work of Smithies et al. (2020) divided the granitic rocks of the Yilgarn into a series of geochemical types. Plotting these up over my Western Yilgarn magnetic interpretation, an immediate association became apparent of a concentration of low-Ca, high-K, high Sr-Y granites (blue-green) parallelling the major crustal boundaries identified from the gravity interpretation and judicious squinting at the gravity and magnetic worms maps.

The hypothesised Koorda Terrane is supported by an axis of potassic monzogranites mirroring the purported boundaries interpreted from gravity and magnetic worms. This mirrors the boundary between the ancient 2.9-3.4Ga Corrigin Tectonic Zone, and the northern extremities of the Wongan Hills and other greenstone belts within the Southwest Terrane, and sutures the southern extents of the Murchison, which is predominantly <2.9Ga.

Additionally, a suite of HFSE and calcic granites occurs north of Perenjori and Morawa, attesting to deep crustal magma sources for granites in this region which do not persist south east through the Koorda axis.

The interesting factor here is that this suite of low-Ca, high-K high Sr/Y granites is also enriched in TREE. This defined a terrane-scale opportunity for REE enriched granites with possibilities to form regolith hosted REE's, some of which I would argue, may form ionic or colloidally bound REE within kaolinised saprolite.

Winnowing: Area Selection

I was taught a mineral exploration methodology which comprised, first and foremost, selecting terranes and tectonic regions, and then selecting camps and districts within them, and eventually driving down into finer scale prospects and finally picking up tenure. This process is often skipped in the modern context in favor of a real estate driven process, whereby owner-occupiers drive deals (often with hidden kickbacks to friendly brokers) based on their tenure in an area, and exploration effort is focused into owner-occupied tenements regardless of whether the tenement is prospective or a logical winnowing process has occurred; the lucky get a discovery and a deal with a major.

Here, as a privateer, I had no competitors at the time and vast acreage to contemplate, and hence a need to undertake a strict area selection process to reduce the tenement pegging to only the most prospective rocks, for an acreage I could afford to apply for in advance of a deal. Here, I relied upon a two-pronged approach; prospectivity modelling and data generation.

Prospectivity Modelling

To construct a prospectivity map, a process was undertaken using a summed grid Weighs of Evidence process, with methodology as below.

This approach was chosen because it is a powerful way of spatially collating prospectivity information in a reasonably bias-free way if one assembles the evidence criteria logically, and one can weight the prospectivity criteria fairly. This is of course somewhat of a personal choice as to the score and rank system and what meat one puts into the sausage machine.

It is also, broadly, what machine learning algorithms do, but allegedly over 155 variables. There is however a point at which one must weigh the falsity of complexity in assigning 155 variables to a process which has only a few key controls; we are often dazzled by overly technical and sophisticated processes when simple things suffice. Anyway, a small rant.

Granite Type

The granite type from the GSWA report 06/2020 (Smithies et al 2022) were assigned a rank from 1 to 9 and the granite class assigned manually to granitoid polygons within the interpretation. Many larger granitoid masses have different assignations of samples within them, so the average composition of these samples was considered and a rank qualitatively assigned. Your mileage may vary here, so take it as my opinion.

Regolith

Regolith development is a key contributor to the presence of a regolith hosted REE deposit, as logically follows. To assess the regolith influence on the prospectivity for regolith REE's, my personal judgement was used to construct a series of scores and a ranking scheme from 1 to 9 based on my perception of the liability of that regolith landform to host a regolith REE deposit.

In essence, I interpret lateritised regolith as prospective, as lateritisation in the Western Yilgarn tends to overly deep saprolitisation. This isn't strictly true, as around Mukinbudin laterite pediments onlap older silcretised caps on granitic regolith or fresh whalebacks, attesting to a complex regime. However, on a terrane scale it certainly holds that laterite reflects a regional development of deep and prolonged weathering, especially given the flanks of the fresh hills tend to host detrital (colluvial) material which weathers in-situ. Examples from Uganda and Tasmania and the Booanya Suite-sourced deposits OD6 has in the eastern Albany-Fraser, show that colluvial and alluvial sediments are candidates for weathering to produce ionic and colloidal REE clays.

Again, YYMV with how you would go about ranking a sizeable portion of the Yilgarn Craton.

The imputed ranking score was then, at great expense to my CPU, assigned to the GSWA 1:500K regolith polygon map in MapInfo on a nominal 800m cell size grid, and the polygon score assigned to a raster grid.

TREO Gridding

Finally, the direct geochemical evidence of TREO contents of granite rocks was gridded.

Geochemistry grids were calculated for three primary and three secondary geochemical indicators of IAC-type REE’s. The three primary indicators are LREO ppm, HREO ppm and F , with secondary indicators of Th, Zr, P2O5%. Primary indicators are the direct measurement of REE’s and fluorine which may correlate with soluble REE phases, whereas secondary indicators are elements typically associated with REE minerals in granites but which are not typically soluble REE’s. These are given a lower ranking score because they may indicate increased prospectivity but do not necessarily do so.

Grid values were assigned via Natural Break of all values >0 divided into 5 classes, with the top class of the primary indicators LREO and HREO assigned a value of 1.5, second 1, third 0.5, fourth 0.25 and the fifth 0.1. Secondary indicators had values of 0.5, 0.25, 0.1, 0.05 and 0.01 respectively, null values 0.1. All non-assayed samples were assigned a score of 0.25 representing an average concentration of the element in the absence of data.

The results speak clearly to supporting the Koorda Axis as a distinct grantic association of TREO enriched high-K, high-P, high-F monzogranites constrained to a distinct magmatic and tectonic terrane between the Southwest Terrane and Yilgarn Craton.

Final Prospectivity Grid

The final prospectivity grid was constructed by summing the gridded values for lithology, regolith and geochemistry, on a notional 800m resolution. This is presented below.

The final product illustrates the Koorda Axis or terrane, as highly prospective, and defined by potassic monzogranites. Areas to the south are less prospective, driven by a more erosional regolith landscape, higher-Ca and lower-P granite types (dominantly TTG's of >2.9Ga). Certain belts within the Southern Cross Domain, with younger c. 2.8-2.7Ga syenitic character, are prospective.

Phase 2: Data Generation

Armed with a prospectivity model defining (at the time) an un-tenured terrane scale REE opportunity equal or better than the Munglinup Gneiss Belt opportunity, several factors caused me to have to work harder prior to taking up tenure. First factor was that in a 350km by 150km area between Wubin, Southern Cross, Koorda and the arse end of the Karroun Plutonic Massif, there were only around 40 samples. This was too few samples to support a tenuring process. So I had to go out on my own bat, at my own expense, and sample more granites. This was done in the summer of 2022.

I sampled a total of 87 rock chips of granites and 350 soils (here I would like to thank Mr Anees Sabet, and Zinc of Ireland, for contributing to this knowledge base). Rock chips were assayed via LITH204 at Intertek Genalysis, and soils via 4-acid with REE add-on. Total outlay was approximately $24,000.

Results showed up to 3,120ppm TREO (including nearly 400ppm HREE) from an amphibole bearing monzosyenite at Mollerin Rock, north of Koorda.

Apatite-Magnetite Monzogranites

The result of this work was that tenure was applied for, but the market for IAC's and regolith hosted REE's passed. The market, or investors, are now more appraised of the various clay projects and the metallurgical barriers to success in Australia are better understood, as are the logistical and operating cost parameters in Western Australia in particular.

However, this work has resulted in an important contribution to understanding the distribution of REE's in granites, and illustrates a first-principles approach to generating province-scale mineral exploration opportunities (two in a row, as has been demonstrated).

To conclude, it is clear that granite rocks in the Yilgarn Craton can be heavily enriched in REEs. Whilst it is only one rock chip sample, the amphibole monzosyenite at Mollerin Rock is at least as elevated in REE's as Venture Minerals' prospect at Brothers, where deep REE enriched regolith consistently returns high grade assays.

There is a distinct distribution of similar granitic rocks throughout the Yilgarn Craton. Venus Metals' prospect at Marvel Loch, in the Southern cross domain, is an example of magnetite-apatite monzogranites of a presumed high-K, high-P character. Here REE appears associated with allanite, a REE-bearing mineral.

Other work, by private concerns, around Mukinbudin (for whom I have consulted ad-hoc) have shown association of REE's with magnetite and ilmenite, as exsolutions and secondary phosphates. Granitoids at Mukinbudin, Kellerberrin and Cunderdin are all of the same Mt-Ap-monzogranite and monzosyenite.

In general the prospective granites form arcuate moderate to intense magnetic anomalies, and are late-stage, coarsely crystalline, and contain primary igneous amphibole and coarse biotite (potentially allanite, though I am not so brave to say so in hand specimen).

High phosphorus contents are likely accounted for by apatite, although monazite is also likely. Fluorine contents are, at times, >900ppm, and up to 0.24%. In general, this attests to a highly evolved mantle volatile input and assists with the fluxing of phosphates into the melt.

Fractionation is not extraordinary for the bulk plutonic masses (it cannot be on a volume basis) so the REE content cannot be explained by silica-mediated fractionation as it is in some highly evolved A-type rhyolites, examples being Kathleen Valley Ignimbrite in the Musgraves (Medlin 2017) Round Top Mountain in Texas (Jowitt et al, 2017; Pingitore et al. 2014). This attests to a source rock control on REE contents, as partial melting systematics alone appears incapable of contributing greatly to REE content (though I admit I'm rusty on my petrogenesis literature).

To my mind the Koorda Terrane represents a suite of granitoids (of various ages) with a unique mantle heterogeneity, and a distinct temporal and spatial association, which is associated with the amalgamation of the Murchison Domain with the Southwest Yilgarn cratonic core.

The distribution of these allanite/apatite monzogranites within Western Australia is shown below; this forms an association with the edges of the western Yilgarn Craton and the >2900Ma crustal architecture. Venture Minerals Limited's Brothers REE prospect is the northern most example of this suite of granites; vast areas of monzogranites and monzosyenites of this association stretch all the way down to VMC's example in the south east.

References

Carbonatites and carbothermalites: A revised classification. Mitchell R.H. & Gittins J., 2022. Lithos 430-431, https://doi.org/10.1016/j.lithos.2022.106861

JOWITT S.M., MEDLIN C.C., CAS R.A.F., 2017. THE RARE EARTH ELEMENT (REE) MINERALISATION POTENTIAL OF HIGHLY FRACTIONATED RHYOLITES: A POTENTIAL LOW-GRADE, BULK TONNAGE SOURCE OF CRITICAL METALS. ORE GEOLOGY REVIEWS 86, P. 548–562.

Lowrey, JR, Smithies, RH and Champion, DC 2022, Yilgarn Granite Project – notes to accompany 2022 data release: Geological Survey of Western Australia, Record 2022/9, 3p.

MEDLIN, CC 2017, THE VOLCANOLOGY, PETROGENESIS, AND ECONOMIC POTENTIAL OF THE MESOPROTEROZOIC SHALLOW-WATER, INTRA-CALDERA, LAVA-LIKE RHEOMORPHIC KATHLEEN IGNIMBRITE, WEST MUSGRAVE PROVINCE, CENTRAL AUSTRALIA: GEOLOGICAL SURVEY OF WESTERN AUSTRALIA, REPORT 171, 294P.

PINGITORE N, CLAGUE J, GORSKI D, 2014. ROUND TOP MOUNTAIN RHYOLITE (TEXAS, USA), A MASSIVE, UNIQUE Y-BEARING-FLUORITE-HOSTED HEAVY RARE EARTH ELEMENT (HREE) DEPOSIT, JOURNAL OF RARE EARTHS, VOLUME 32, ISSUE 1, 2014, PAGES 90-96.