Superhydride Synthesis Study by Machine Learning Potential Molecular Dynamics

Room-temperature superconductivity is one of the foremost goals in physics. Superhydrides, with critical temperatures up to 260 K, have emerged as promising candidates. They are typically synthesized in diamond anvil cells (DACs) under high pressure and high temperature via laser heating, indicating kinetically slow reactions with large activation barriers. Gaining physical insight into these processes is essential for developing efficient synthesis routes.

Here, we directly observe superhydride formation using machine-learning potential molecular dynamics simulations. We show that CaH₂ in contact with dense H₂ undergoes surface melting, followed by hydrogen uptake into a disordered phase. This produces a metastable, liquid-like CaH₄ intermediate that subsequently crystallizes into solid CaH₄. The reaction is thermodynamically driven by hydrogenation enthalpy, which lowers the barrier for surface melting. Importantly, the pathway is favored under “overpressure,” where the applied pressure exceeds the equilibrium pressure predicted by density functional theory. In addition, we will also present our latest results on CaH₆ synthesis.