Revealing the molecular structures of α-Al2O3(0001)-water interface by machine learning based computational...

Area of Science:

• Surface science and spectroscopy
• Computational chemistry and materials science
• Machine learning applications in physical sciences

Background:

• Solid-water interfaces are critical in physical and chemical processes.
• Surface-specific sum-frequency generation (SFG) spectroscopy is key for studying these interfaces.
• Accurate theoretical calculations, often using molecular dynamics (MD) simulations, are needed to interpret SFG spectra, but computationally expensive ab initio MD (AIMD) is limited by long trajectory requirements.

Purpose of the Study:

• To develop and demonstrate machine learning (ML)-accelerated methods for calculating vibrational spectra (IR, Raman, SFG) of solid-water interfaces.
• To overcome the computational cost and time limitations of traditional AIMD simulations for complex interfaces.
• To enable more rapid and efficient analysis of interfacial water structures.

Main Methods:

• Utilized machine learning (ML) to accelerate ab initio molecular dynamics (AIMD) simulations.
• Employed both dipole moment-polarizability correlation function and velocity-velocity correlation function approaches for SFG spectra calculation.
• Calculated vibrational spectra including IR, Raman, and SFG for solid-water interfaces.

Main Results:

• Successfully accelerated AIMD simulations using ML methods.
• Demonstrated the capability to calculate SFG spectra of solid-water interfaces with ML acceleration.
• Achieved faster and lower-cost computation of vibrational spectra compared to conventional methods.

Conclusions:

• Machine learning provides a viable pathway to accelerate the calculation of vibrational spectra for solid-water interfaces.
• This ML-accelerated approach significantly reduces computational cost and time, making complex systems more accessible for study.
• The developed methods offer a powerful tool for understanding interfacial water structures through SFG spectroscopy.