Data-driven discovery and rapid, direct synthesis of MXenes

MXenes, two-dimensional transition-metal carbides and nitrides, are typically obtained from MAX phases, yet historical reports suggest a broader, largely unexplored chemical space. Here we combine machine-learning-assisted database mining with experiments to uncover overlooked multilayer (ml) MXenes. Screening of repositories reveals a "Treasure Chest" of 38 previously synthesized but unrecognized ml-MXene candidates. Guided by these findings, we rediscover five MXenes using a rapid, scalable self-propagating high-temperature synthesis that requires no sustained external heating and completes within minutes. Inspired by the identified chemistries, we further realize 11 previously unexplored rare-earth-based M2CT2 MXenes (M= Pr, Nd, Sm, Gd, Tb, Ho, and Tm). Experiments and theory reveal semiconducting behavior and diverse magnetic states across this family. Together, these results expand the MXene family and demonstrate a data-driven strategy for accelerating materials discovery through sustainable methods.