Investigation on Melting Curves and Phase Diagrams for CaO₃ Using Deep Learning Potentials

Area of Science:

• Planetary Science
• Geophysics
• Materials Science

Background:

• Deep planetary melting influences core formation, magnetic fields, and crustal evolution.
• Understanding planetary interiors requires knowledge of deep geological processes.

Purpose of the Study:

• Investigate melting curves and P-T phase diagrams for CaO3, a candidate lower mantle mineral.
• Utilize a deep learning potential model for accurate simulations.

Main Methods:

• Employed first-principles, molecular dynamics, and quasi-harmonic approximation.
• Verified deep learning potential model using equations of state and phase transition pressures.
• Calculated melting temperatures using single-phase, void, and two-phase methods.

Main Results:

• Determined zero-pressure melting temperatures of 975 K (single-phase), 850 K (void), and 755 K (two-phase).
• Analyzed melting processes via radial distribution function and mean-square displacement.
• Generated CaO3 melting phase diagrams up to 135 GPa using the two-phase method.

Conclusions:

• The deep learning potential model accurately simulates CaO3 behavior under high pressure and temperature.
• This research provides crucial data for understanding lower mantle composition and dynamics.
• Results contribute to models of planetary evolution and interior processes.