Data-Driven Discovery of Sulfur Mineralogy

Mineralogy is entering a new era of data-driven discovery. Sulfur (S) plays a pivotal role in global biogeochemical cycles and Earth’s redox evolution. Sulfur-bearing minerals provide key records of the local environmental conditions during their formation, reflecting the chemical speciation and redox states. In this study, we employ advanced data-driven methods to analyze 239,544 mineral–locality pairs from the Mineral Evolution Database (MED; RRUFF.info/Evolution), encompassing 1,028 sulfur mineral species, with 71,622 of these pairs including age information. This comprehensive investigation examines the temporal evolution, diversity, and spatial distribution patterns of sulfur minerals throughout deep geological time. Our findings reveal that sulfur mineralization exhibits episodic growth patterns that correlate with five supercontinent assembly events. The redox state signatures preserved in sulfur minerals offer critical insights into the evolutionary trajectory of Earth’s crust. The data further indicate that crustal oxidation, driven by atmospheric processes, progresses as a surface-to-depth phenomenon, transitioning from shallow to deeper crustal levels. In our examination of sulfur mineral ecology, we implemented the Poisson-lognormal Large Number of Rare Events (LNRE) model, predicting a minimum of 2,476 sulfur mineral species existing in Earth’s crust. This estimation suggests that 1,448 species remain undiscovered, a conservative figure considering sampling limitations and potential technological breakthroughs. Application of the LNRE model to sulfur mineral subsets indicates that the majority of undiscovered species are likely sulfides and sulfosalts formed in endogenic reducing environments. Network analysis of sulfur minerals reveals distinct partitioning between S–O bonded and non-S–O bonded species, reflecting fundamental trends in oxygen fugacity. Community detection within the network of the 402 most common sulfur minerals identified four distinct clusters, each representing unique paragenetic environments. The characteristic node distribution patterns observed in sulfur mineral bipartite network diagram further validate the applicability of the LNRE model to sulfur mineral distribution patterns.