Lilia Boeri

We present a method for efficiently predicting approximate binary convex hulls through the automatic generation of a minimal basis set of representative structural templates. The approach is based on the assumption that chemical space can be partitioned into a limited number of regions with similar bonding characteristics, such that a compact set of templates can capture the key structural and energetic trends across a wide range of compositions. Templates are selected via a data-driven algorithm informed by evolutionary structure searches, and subsequently reused to estimate formation enthalpies across diverse systems using limited DFT relaxations. We validate the method with a comprehensive benchmark dataset of 105 binary element pairs. Convex hulls predicted by our method are compared to evolutionary algorithm predictions using multiple metrics, including composition overlap, hull depth error, structural agreement. When targeting an average enthalpy error of 0.1 eV/atom, sufficient for high-throughput thermodynamic screening—the method achieves a 15-fold reduction in the number of DFT relaxations while maintaining 87% accuracy in identifying stable compositions. The approach is designed for scalability and applications in high-throughput exploration of complex material spaces.

We identify a family of boron-rich compounds consisting of interconnected B 12 icosahedra and electropositive guest atoms ( X ) in interstitial sites. These structures were found through first-principles crystal structure prediction at 50 GPa, and are dynamically stable down to ambient pressure. When X is a mono- or trivalent element, the structures are metallic and superconducting. Predicted critical temperatures reach up to 42 K for CsB 12 , rivaling MgB 2 , the highest- T c ambient-pressure conventional superconductor. We interpret the X B 12 phase as a superatomic crystal: the B 12 units retain the icosahedral shape that they also exhibit in isolation, while forming an extended crystalline network. When X is a mono- or trivalent atom, the system is metallic, and the B–B covalent bonding promotes strong electron-phonon coupling. Unlike MgB 2 , where superconductivity is driven by a narrow subset of phonon modes, the X B 12 compounds exhibit broad, mode- and momentum-distributed coupling through both intra- and intersuperatomic vibrations. Our results highlight the X B 12 family as a promising platform for superconductivity and demonstrate the potential of superatoms as functional building blocks in solid-state materials design.

We critically reexamine the superconducting properties of rock-salt transition-metal carbides (TMCs), often regarded as textbook conventional superconductors, combining first-principles electron-phonon calculations with variable-composition evolutionary structure prediction. Studying superconducting trends across the entire transition-metal series, we find that, when the rock-salt stoichiometric phase is dynamically or thermodynamically unstable, carbon-vacant structures identified through unbiased structure prediction permit to reconcile theoretical calculations with experimental trends. Our integrated use of structure prediction and electron-phonon calculations defines a general framework for realistic modeling of superconductors shaped by non-equilibrium synthesis routes and defect tolerance.

Nb 3 Sn is the workhorse superconductor for high-field applications; however, several fundamental aspects remain open 70 years after its discovery. Using anharmonic Migdal-Eliashberg theory, we resolve three key questions: (1) the cubic-to-tetragonal martensitic transition is weakly first order; (2) the superconducting gap is strongly anisotropic but continuous over the Fermi surface; and (3) the drop of H c 2 across the transition arises from both reduced electron-phonon coupling and a redistribution of Fermi velocities. The orbital origin of the gap anisotropy suggests targeted doping strategies: Sn -site substitution can increase both T c and H c 2 , while Nb -site substitution raises H c 2 at the cost of lowering T c .

We critically reexamine the superconducting properties of rock-salt transition-metal carbides (TMCs), often regarded as textbook conventional superconductors, combining first-principles electron-phonon calculations with variable-composition evolutionary structure prediction. Studying superconducting trends across the entire transition-metal series, we find that, when the rock-salt stoichiometric phase is dynamically or thermodynamically unstable, carbon-vacant structures identified through unbiased structure prediction permit to reconcile theoretical calculations with experimental trends. Our integrated use of structure prediction and electron-phonon calculations defines a general framework for realistic modelling of superconductors shaped by non-equilibrium synthesis routes and defect tolerance.