Thèse de l'Espace au Terrain Développement d'Une Méthodologie d'Exploration Multi-Échelle des Terres Rares dans les Systèmes à Carbonatites Combinant les Données Hyperspectrales Enmap et la Fluore H/F Doctorat.Gouv.Fr

Établissement : Université de Toulouse École doctorale : SDU2E - Sciences de l'Univers, de l'Environnement et de l'Espace Laboratoire de recherche : GET - Geosciences Environnement Toulouse Direction de la thèse : David BARATOUX ORCID 0000000217855262 Début de la thèse : 2027-09-01 Date limite de candidature : 2026-11-23T23:59:59 Les éléments des terres rares (REE) sont des matières premières critiques essentielles à de nombreuses technologies de la transition énergétique, notamment aux aimants permanents utilisés dans les véhicules électriques et les éoliennes. Les complexes carbonatitiques et alcalins abritent certaines des plus importantes ressources mondiales en REE et constituent donc des environnements majeurs pour l'exploration. Cependant, leur exploration reste difficile en raison de leur forte hétérogénéité minéralogique et géochimique. Les concentrations en REE peuvent varier de quelques centaines de ppm à plusieurs pourcents sur de courtes distances, tandis que les signatures primaires peuvent être fortement modifiées par l'altération.

L'altération peut modifier profondément la distribution des REE dans les systèmes carbonatitiques. La dissolution des carbonates et d'autres minéraux primaires peut conduire à la formation d'un régolithe résiduel où les REE peuvent être fortement concentrés ou redistribués. La plus forte anomalie géochimique de surface peut ainsi ne pas se situer directement à l'aplomb de la zone minéralisée primaire. Comprendre les échelles spatiales de cette redistribution est essentiel à l'exploration. Ce projet vise à déterminer comment les REE sont redistribués au cours de l'altération des complexes alcalins à carbonatites, à caractériser les contrôles minéralogiques, géochimiques et environnementaux de leur organisation spatiale, et à établir comment ces processus influencent leur détectabilité par télédétection hyperspectrale. Une attention particulière sera portée aux échelles spatiales caractéristiques de l'enrichissement en REE et à distinguer les signatures spectrales directes des minéraux porteurs de REE des signatures indirectes des assemblages d'altération.

Le projet portera sur deux systèmes carbonatitiques : le complexe de Twihinate, dans le sud du Maroc, situé dans un environnement aride, et le complexe de Monte Muambe, au Mozambique, caractérisé par une altération tropicale intense et un régolithe développé. Cette comparaison permettra d'étudier l'influence de la minéralogie primaire, de l'altération météorique et des conditions de surface sur la redistribution des REE et d'évaluer les limites de leur détection hyperspectrale.

Le projet utilisera les données hyperspectrales acquises par la mission EnMAP, opérée par le Centre allemand pour l'aéronautique et l'astronautique (DLR). EnMAP fournit environ 230 bandes spectrales couvrant les domaines VNIR et SWIR, avec une résolution spatiale de 30 m, et des produits de niveau 2A corrigés des effets atmosphériques, accessibles au public. La spectroscopie de laboratoire et de terrain caractérisera les signatures des minéraux porteurs de REE, notamment des phases riches en Nd. Leur composition géochimique et minéralogique sera déterminée par ICP-MS/ICP-OES, DRX et techniques complémentaires pour constituer des références d'interprétation des observations orbitales. Des mesures géochimiques de terrain à haute densité, réalisées à l'aide de la fluorescence X portable (pXRF), permettront de caractériser les éléments majeurs et certains éléments traces utilisés comme indicateurs géochimiques, pour documenter la continuité spatiale et l'hétérogénéité de l'enrichissement en REE. Ces données permettront de quantifier les échelles spatiales de l'enrichissement et la continuité nécessaire à sa détection satellitaire.

La thèse aboutira au développement d'une méthode d'exploration multi-échelle « space-to-field », intégrant les informations géologiques et minéralogiques, la géochimie de terrain, la spectroscopie et les observations EnMAP. Le projet apportera aussi de nouvelles connaissances sur les processus contrôlant l'enrichissement, la redistribution et l'organisation spatiale des REE dans les carbonatites altérées. Les résultats seront également pertinents pour l'exploration planétaire, en fournissant un cadre pour l'interprétation des données spectrales orbitales. Rare earth elements (REE; lanthanides and yttrium, atomic numbers 57-71 and 39) are strategic metals essential for modern technologies, including renewable energy systems, electric mobility, and advanced electronics. Neodymium (Nd), for example, is a key component of high-performance NdFeB permanent magnets used in electric vehicles and wind turbines. Due to their economic importance and supply risks, REE are classified as critical raw materials (CRMs) under the European Union Critical Raw Materials Act (European Commission, 2023).

Carbonatite-related systems host some of the world's largest and highest-grade REE deposits and represent one of the most important geological environments for REE mineralization (Wang et al., 2020). Carbonatites are rare igneous rocks containing more than 50 vol% carbonate minerals and generally less than 20 wt% SiO. Unlike most silicate rocks, they may contain several wt% REE, although economically significant mineralization is generally restricted to specific lithological facies where REE are concentrated in minerals such as carbonates (e.g., bastnäsite, synchysite, parisite) and phosphates (e.g., monazite, xenotime, and fluorapatite).

Carbonatite systems display a remarkable range of spatial scales, from REE-bearing mineral grains and aggregates (approximately 0.01 - 10 mm), through ore bodies and high-grade zones (1 - 1000 m) to carbonatite complexes extending over several kilometers and associated alkaline provinces extending over tens to hundreds of kilometers (e.g., Walter et al., 2026). At each of these scales, REE distribution is strongly heterogeneous, reflecting variations in carbonatite lithology, mineral assemblages, the abundance and distribution of REE-bearing minerals, and subsequent magmatic, hydrothermal, and weathering processes. A well-known example is the Mountain Pass deposit (USA), where bastnäsite-rich carbonatite facies locally contain very high REE concentrations, whereas other facies are considerably less enriched (Verplanck et al., 2014). This multi-scale heterogeneity is a fundamental characteristic of carbonatite-related REE systems and represents a major challenge for their characterization and exploration.

The exploration of carbonatite-related REE systems is particularly challenging because carbonatites commonly constitute only a small component of larger alkaline complexes dominated by silicate rocks. Their identification and delineation can therefore be difficult, particularly where exposure is limited. Moreover, carbonatites are highly susceptible to weathering, which can profoundly modify their primary mineral assemblages and redistribute REE through mineral dissolution, leaching, transport, and formation of secondary mineral (e.g., de Oliveira, 2026). Depending on the mineralogical and hydrological conditions, weathering may preserve part of the geochemical signature of the underlying carbonatite, redistribute REE within the regolith, or generate secondary REE enrichment and supergene mineralization. Weathering therefore affects not only REE concentrations and mineralogy but also their spatial organization at the surface.

This is particularly important for remote sensing because hyperspectral satellites principally characterize exposed near-surface material. The spatial scale at which an REE-bearing system can be detected from orbit may therefore depend strongly on the degree of regolith development, surface exposure, vegetation cover, mineralogical composition, and the spatial continuity of REE-bearing minerals. The characteristic spatial correlation or heterogeneity of REE concentrations may consequently differ substantially between contrasting climatic and weathering environments, such as the relatively arid regions of Morocco and the humid tropical environments of Mozambique. Understanding how REE concentrations and mineralogical signatures are organized across these different spatial scales is therefore essential for developing effective remote-sensing-based exploration strategies.

Current REE exploration relies mainly on geological mapping, systematic sampling, drilling, and laboratory-based geochemical analyses using techniques such as ICP-MS and ICP-OES. Although these methods provide accurate measurements, they are expensive and time-consuming and are difficult to deploy at high sampling density over large or remote areas. Moreover, conventional exploration datasets generally provide limited information on the multi-scale architecture of REE mineralization in carbonatite systems. Developing rapid and cost-effective approaches capable of identifying prospective zones before detailed exploration is therefore a major scientific and industrial priority. Where applicable, remote sensing offers a cost-effective approach during the early stages of mineral exploration, because of its capacity to assess large portions of the Earth's surface.

Recent advances in orbital hyperspectral remote sensing have opened new opportunities for CRM exploration (Asadzadeh & Chabrillat, 2025). The Environmental Mapping and Analysis Program (EnMAP), launched in 2022 by the German Aerospace Center (DLR), provides freely available hyperspectral imagery covering the visible and near-infrared (VNIR; 400-1000 nm) and short-wave infrared (SWIR; 1000-2500 nm) regions. Its spectral characteristics are suitable for identifying diagnostic absorption features associated with REE-bearing minerals (Möller & Williams-Jones, 2018; Asadzadeh et al., 2024; Kumar et al., 2026; Asadzadeh & Chabrillat, 2025; Neave et al., 2016; Gadea et al., 2024; Boesche et al., 2015; Zimmermann et al., 2016; Booysen et al., 2019). The detection of REE from hyperspectral data is primarily based on characteristic electronic absorption features of REE ions (e.g., Neave et al., 2016). Several REE ions have diagnostic absorption features in the VNIR, but Nd³ is particularly useful because it produces relatively strong and distinctive absorption features around 580-600 nm, 740-800 nm and 850-900 nm (Fig. 1, cf. Complementary form). Their exact position, intensity, and spectral expression depend on Nd concentration, mineralogical host, crystal chemistry and grain-scale distribution. These features have been investigated in laboratory spectra (e.g., Möller & Williams-Jones, 2018) and, more recently, using airborne (e.g., Kumar et al., 2026) and orbital hyperspectral observations (e.g., Asadzadeh et al., 2024). Recent studies at the Mountain Pass carbonatite (USA) (Asadzadeh et al. 2024) and alkaline complexes in India (Kumar et al., 2026) have demonstrated the potential of EnMAP for detecting REE-related, particularly Nd-related, spectral responses at regional scales. In summary, EnMAP has sufficient spectral sampling (~6.5 nm in the VNIR) and resolution (~12 nm in the VNIR) to detect the strongest diagnostic spectral features of Nd, but their detectability depends on mineralogy, abundance, spatial continuity, surface mixing and environmental conditions. The influence of lithology, weathering intensity, regolith development, ferric mineral abundance, vegetation cover, grain size, and the spatial distribution of REE-bearing minerals on satellite-scale detection remains poorly constrained. This knowledge gap currently limits the transferability of hyperspectral approaches between different geological and climatic environments.

Field- and laboratory-based spectroscopy can provide an essential link between mineral-scale processes and orbital observations. VNIR/SWIR measurements of representative fresh-rock and regolith samples under controlled laboratory conditions will provide reference spectra for REE-bearing minerals and their host materials. These measurements will allow the relationships between mineralogy, REE concentration, weathering, and spectral response to be established before comparison with field and satellite observations. Field-based VNIR/SWIR measurements will further assess how these spectral signatures are expressed under natural surface conditions.
Field-based geochemistry will provide an additional intermediate scale between orbital observations and laboratory analyses. Handheld (pXRF) measurements offer a rapid and non-destructive means of acquiring dense geochemical datasets across large numbers of field samples. Their high sampling density can be adapted to the spatial variability of carbonatite systems, allowing short-scale chemical heterogeneities, lithological contrasts and geochemical gradients to be mapped. In this context, pXRF can provide an important link between detailed field observations and the coarser spatial scale of hyperspectral satellite imagery.

Although pXRF is widely used in mineral exploration, its application to direct REE quantification remains challenging because of matrix effects, mineralogical heterogeneity, and the relatively low abundance of many REE in natural materials (Coe et al., 2026; Akhmetzhanov et al., 2025). Rather than replacing laboratory-based REE analyses, pXRF will therefore be investigated as a complementary high-density geochemical tool, with its performance and applicability assessed through calibration against high-quality laboratory datasets. Particular attention will be given to elements that can characterize lithological and mineralogical variations and help identify zones associated with REE enrichment. Recent analytical developments, including our ongoing work using micro-XRF (Pourkhorsandi et al., 2023), indicate that optimized XRF-based approaches can provide valuable information for the rapid characterization of REE-rich carbonatites and their weathering products. Developing robust field protocols based on calibration with high-quality laboratory datasets could therefore improve the efficiency of early-stage REE exploration.

This PhD project proposes to integrate field-based geochemistry, laboratory mineralogical and geochemical characterization, laboratory and field VNIR/SWIR spectroscopy, and EnMAP hyperspectral observations to establish a validated space-to-field exploration workflow for REE in carbonatite systems. Laboratory spectroscopy will provide the mineralogical and spectral reference framework, field measurements will characterize spatial variability at the meter-to-centimeter scale, and EnMAP observations will enable regional-scale targeting. The comparison of contrasting carbonatite systems in Morocco (Twihinate) and Mozambique (Monte Muambe) will provide a framework to determine how geology, mineralogy, weathering, and surface conditions control REE distribution and remote detectability, while contributing to improved exploration strategies for CRMs.
The objectives of this project are twofold. The first objective is to investigate the geological, mineralogical, and environmental controls governing rare earth element (REE) distribution within carbonatite-bearing alkaline complexes. Particular attention will be given to the relationships between REE enrichment, carbonatite lithology, mineralogy, and weathering processes. Two contrasting carbonatite systems will be investigated: the Monte Muambe carbonatite complex in Mozambique (Lehto & Gonçalves, 2008), developed under tropical weathering conditions with extensive regolith formation, and the Twihinate carbonatite complex in southern Morocco, located in the arid Saharan environment, with limited present-day weathering and sparse vegetation (Bouabdellah et al., 2022; Boukirou et al., 2022; Boukirou et al., 2023). These carbonatites were selected based on existing projects conducted by the PI and co-workers, who have facilitated access to the study areas. Their comparison will provide a unique framework to evaluate how climate, regolith development, mineralogical variability, iron-oxide abundance, and surface conditions, including vegetation cover, influence REE redistribution and their detectability by remote sensing.

The second objective is to develop an integrated, rapid, multi-scale and cost-effective exploration workflow combining handheld X-ray fluorescence (pXRF) (Coe et al., 2026) and laboratory- and field-based visible and near-infrared (VNIR) and short-wave infrared (SWIR) spectroscopy (Möller & Williams-Jones, 2018), and EnMAP hyperspectral satellite observations (Asadzadeh et al., 2024; Kumar et al., 2026). Geological mapping, laboratory geochemistry, mineralogical characterization, laboratory and field VNIR/SWIR spectroscopy, field-based pXRF measurements, and orbital hyperspectral data will be integrated to establish and validate a methodology for REE exploration applicable to carbonatite systems. A key scientific question will be addressed: at which spatial scales do REE concentrations vary in carbonatite systems, and which of these scales can be resolved by hyperspectral satellite observations? Determining the characteristic spatial correlation scales of REE enrichment and their relationship with spectral responses will help identify the factors controlling the success and limitations of hyperspectral detection of REE mineralization. Together, these objectives will contribute to the development of next-generation exploration strategies for critical raw materials (CRM).
The project combines field investigations, laboratory characterization, laboratory and field VNIR/SWIR spectroscopy, handheld X-ray fluorescence (pXRF) measurements, and orbital hyperspectral observations to develop an integrated workflow for REE exploration in carbonatite systems. The methodology aims to determine the characteristic spatial scales of REE variability and understand how these scales affect the detectability of mineralization by hyperspectral satellite observations. The Monte Muambe and Twihinate sites represent complementary natural laboratories for investigating how contrasting climatic conditions, weathering intensity, regolith development, and surface characteristics influence REE redistribution and their detectability from hyperspectral data. The following hypotheses will be tested:

1. Weathering modifies both REE concentrations and their spatial distribution, producing different characteristic patterns of REE enrichment in tropical and arid regolith.
2. The detectability of REE mineralization by EnMAP depends not only on REE concentration but also on the spatial continuity and abundance of REE-rich domains.
3. The characteristic spatial scales of REE variability differ between tropical and arid environments and control the detectability of REE enrichment at the 30 m spatial resolution of EnMAP.

The methodology is structured into five complementary work packages.

WP1. Geological, geochemical, mineralogical, and spectral characterization. Fieldwork will be jointly conducted by the two institutions and their local partners (UMP6, Altona Rare Earths Ltd.). The University of Namur has an extensive network and recently published on the Twihinate carbonatite, while GET has privileged access to the Monte Muambe carbonatite. Field campaigns will include geological mapping, characterization of carbonatite facies and weathering profiles, collection of representative fresh-rock and regolith samples (including soil), and acquisition of in situ pXRF measurements (using instruments available both in GET and University of Namur), and where appropriate VNIR/SWIR measurements. Representative samples will be analyzed for major and trace elements using ICP-OES and ICP-MS at GET, and for bulk-rock and clay mineralogy using X-ray diffraction (XRD) at the University of Namur, complemented where necessary by microscopic observations. Representative fresh-rock and regolith samples will also be measured using laboratory VNIR/SWIR ASD FieldSpec spectrometer available in Toulouse (OMP). These measurements will establish reference spectra for REE-bearing minerals and host materials and allow spectral responses to be related to mineralogy, REE concentration, grain size, and weathering.
Together, these datasets will establish the geological framework and quantify the relationships between lithology, mineralogy, weathering processes, REE distribution, and spectral response.

WP2. Development and calibration of pXRF approaches for 'rapid' REE assessment. Laboratory-based pXRF calibration experiments will use reference materials and samples independently characterized by ICP-MS and ICP-OES, including materials from the University of Namur and the existing GET sample database. Commercial pXRF calibrations generally do not provide robust quantification of REE. Among the REE, La and Ce are expected to be the most readily detected because of their relatively high concentrations in many carbonatites, while Nd may be detectable in strongly enriched materials. Heavy REE are generally more challenging because of lower concentrations and spectral interferences.
Laboratory-based pXRF calibration experiments will therefore be developed using reference materials and samples independently characterized by ICP-MS and ICP-OES. These experiments will evaluate which REE and associated elements can be robustly quantified, together with their detection limits, as a function of concentration, matrix effects, mineralogical variability, major element composition, and weathering. The objective is to establish a robust protocol for rapid field characterization of REE-enriched zones and to generate high-density geochemical datasets suitable for comparison with hyperspectral observation for further calibration.

WP3. EnMAP hyperspectral analysis and spectral characterization.
This WP will be largely supported by the PI at GET, given his extensive experience on remote sensing to support and train the PhD student. EnMAP imagery will be processed to identify spectral responses associated with REE-bearing minerals, with particular attention to Nd-bearing phases and REE carbonates. Laboratory VNIR/SWIR measurements of representative samples will provide the fundamental spectral reference for interpreting orbital observations and assessing how mineralogy, REE concentration, grain size, weathering, and surface conditions affect spectral detectability. These laboratory spectra will be complemented by field-based VNIR/SWIR measurements to assess how spectral signatures are expressed under natural conditions. EnMAP observations will be interpreted using laboratory and field spectral measurements, mineralogical data, reference spectral libraries (e.g., USGS), field-based pXRF measurements, and laboratory ICP-MS analyses. Because EnMAP captures diagnostic Nd absorption features in the VNIR, while pXRF provides complementary elemental information where concentrations permit, these datasets will establish a robust link between orbital observations and ground geochemistry. The resulting spectral analyses will be used to produce REE prospectivity maps and to evaluate the capability and limitations of EnMAP for REE exploration under contrasting geological and climatic conditions.

WP4. Integration of geochemical, mineralogical, spectral, and hyperspectral datasets.
Global integration of field-based and space-based methods will require close collaboration and regular discussions between the two institutions to ensure a robust integration of all data. Geological, mineralogical, geochemical, laboratory spectral, field spectral, and hyperspectral datasets will be integrated to investigate the controls governing the detectability of REE mineralization from orbital observations. Particular emphasis will be placed on the spatial organization of REE mineralization across scales, from field measurements to satellite observations.

High-density pXRF datasets, calibrated using ICP-MS analyses, will be combined with EnMAP imagery to characterize the spatial variability of REE and associated proxy elements. Two-dimensional spatial statistical analyses, including frequency distributions, empirical variograms, cross-variograms, and spatial covariance analyses between REE, major elements, and Nd spectral indices, will be used to quantify characteristic autocorrelation and cross-correlation lengths. This approach will determine how the spatial structure of REE mineralization influences its detectability at the 30 m spatial resolution of EnMAP and help identify the minimum spatial continuity required for reliable satellite detection. Comparison between the Moroccan and Mozambican carbonatite systems will further distinguish the respective influences of primary geological controls (lithology, mineralogy, structural organization) and secondary environmental processes (weathering, regolith development, ferric mineral abundance, and vegetation cover) on REE redistribution and hyperspectral detectability.

WP5. Development of an integrated space-to-field exploration workflow. Both institutions will contribute to the development of this methodology. The results obtained from WP3 and WP4 will determine the characteristic spatial scales of REE enrichment in carbonatite systems and how these scales determine the detectability of mineralization by orbital hyperspectral sensors. Combined with detailed geological, petrological, mineralogical, geochemical, and spectral characterization, these results will lead to the development of an innovative exploration methodology applicable to carbonatite-related REE systems. The project will also identify the strengths, limitations, and uncertainties associated with current approaches and propose recommendations for future exploration strategies and hyperspectral missions dedicated to CRMs.