Soutenance Bence Horányi

Lithium is a high-demand critical metal that is widely used in modern-day technologies that are required for a global energy transition. Approximately 50% of the global lithium supply is sourced from rare-metal granites and pegmatites (RMGPs). In order to meet the rising demand for lithium, it is imperative to discover new ore deposits, which requires precise petrogenetic constraints on the formation of RMGPs. However, despite the economic significance of granitic pegmatites, the mechanism of lithium enrichment in economic-grade deposits (>5000 ppm lithium) relative to conventional granitic rocks in the crust (<100 ppm lithium) remains widely debated. The two competing petrogenetic models of RMGPs include either the high-degree fractional crystallisation of granitic melts or the partial melting of enriched metasedimentary rocks. The degree of lithium enrichment during these processes remains debated due to a lack of precise constraints on the partitioning of lithium between minerals and felsic melts.

In order to better constrain the partitioning behaviour of lithium during magmatic processes, partial melting and crystallisation experiments were performed on metasedimentary rocks and felsic glasses, respectively. Partition coefficients were determined between minerals (micas, feldspars, quartz, staurolite, cordierite, garnet, and tourmaline) and felsic melts that contain 50-6000 ppm lithium. To further constrain the partitioning behaviour of lithium, experimental partition coefficients were compared with the distribution of rare metals in the Richemont rhyolite, which is regarded as the extrusive equivalent of RMGPs. Newly-obtained partition coefficients from the experiments and the Richemont rhyolite vary by up to two orders of magnitude. This variation was parameterised using empirical formulae to capture the evolution of partition coefficients as a function of experimental conditions and the compositions of minerals and melts. The parameterised partition coefficients are consistent with lithium becoming compatible in micas, but increasingly incompatible in feldspars and quartz during magmatic differentiation.

To provide a theoretical framework for the modelled evolution of partition coefficients as a function of melt composition, the effect of lithium on the structure of pegmatite-forming melts was investigated by nuclear magnetic resonance (NMR) spectroscopy. NMR spectra were obtained from natural and synthetic felsic glasses containing up to 2 wt% lithium. The results are consistent with the charge compensation of tetrahedral aluminium by lithium, which shifts the molecular structure of the melts from peraluminous to peralkaline during the progressive enrichment of lithium. The transition in melt composition during magmatic differentiation results in lithium becoming increasingly incompatible in quartz and to a lesser extent in feldspars.

To constrain the petrogenesis of RMGPs, trace element modelling was performed using newly-obtained partition coefficients to simulate the enrichment of lithium during crustal anatexis and fractional crystallisation. Partition coefficients were incrementally varied to capture the evolving compositions of minerals and melts as a function of temperature. The results from the modelling are consistent with rare-metal granites and pegmatites forming in a two-stage process, involving the partial melting of enriched crustal rocks (>300 ppm lithium), followed by the moderate fractional crystallisation of the extracted melts (>1500 ppm lithium). The pre-enrichment of anatectic melts results in lithium becoming highly incompatible in quartz and feldspars, which induces a runaway effect, wherein extracted melts can become efficiently enriched during protracted fractional crystallisation to produce RMGPs. To further constrain petrogenetic models, future studies must combine the use of experimental partition coefficients and robust thermodynamic phase equilibria in low-temperature felsic systems that are enriched in lithium.