Magnesium-ion batteries (MBs) have the potential to revolutionize next-generation energy storage due to the earth-abundant and dendrite-free nature of magnesium, improved safety characteristics, and reduced environmental impact compared with lithium-ion batteries. However, the development of stable solid-state electrolytes remains challenging because thermodynamic phase instability, such as phase separation or miscibility gaps under operating conditions, can lead to structural degradation that compromises both ionic conductivity and mechanical integrity. In this work, the thermodynamic phase stability and configurational energetics of the selenium-doped MgSc 2 S 1-x Se x spinel system, a promising structural framework for potential magnesium solid-state electrolyte materials, are investigated using the semi-empirical Universal Cluster Expansion (UNCLE) approach combined with Monte Carlo simulations. The cluster expansion model predicted 97 candidate configurations across the S-Se compositional range. Analysis of the calculated ground-state convex hull reveals that most intermediate configurations exhibit positive heats of formation, indicating the presence of miscibility gaps across much of the compositional range. However, several selenium-rich compositions are thermodynamically stable at 0 K, with MgSc 2 S 0.25 Se 3.75 identified as the lowest-energy configuration. Finite-temperature behaviour was further examined using Monte Carlo simulations to estimate critical phase transition temperatures for various S-Se compositions. The simulations show that the system tends toward phase separation at 0 K but transitions to a mixed phase at elevated temperatures, with phase transitions occurring within the approximate temperature range of 250 – 400 K depending on composition. These results provide fundamental thermodynamic insights that may guide the design of structurally stable Mg-ion conducting frameworks and help mitigate thermodynamic phase instability issues in solid-state electrolytes. While crucial for overall electrolyte performance, a detailed investigation of ionic conductivity, which typically requires atomistic simulations such as molecular dynamics, is beyond the scope of the present study and will be addressed in future work.
Tibane et al. (Sun,) studied this question.