Mechanistic Insights into Water Autoionization through Metadynamics Simulation Enhanced by Machine Learning.

Area of Science:

• Physical Chemistry
• Computational Chemistry
• Chemical Physics

Background:

• Characterizing the free energy landscape of water ionization is challenging due to computational costs and rare-event dynamics.
• A lack of equilibrium sampling hinders mechanistic understanding of water autoionization.

Purpose of the Study:

• To elucidate the free energy landscape and mechanism of water autoionization.
• To overcome limitations of expensive ab initio calculations and rare-event sampling in studying water ionization.

Main Methods:

• Utilizing nanosecond timescale metadynamics simulations with classical nuclei.
• Employing atomic neural network potentials to enhance simulation accuracy and efficiency.
• Performing conditional ensemble average analyses to probe reaction mechanisms.

Main Results:

• Achieved convergent free energy surfaces for water ionization.
• Accurately reproduced the equilibrium constant (pK_{w}=14.14) and ionization rate constant (1.369×10^{-3} s^{-1}).
• Unveiled a triple-proton transfer at the transition state, occurring via a concerted but asynchronous mechanism.

Conclusions:

• Established a dual-presolvation mechanism involving hypercoordinated and undercoordinated water molecules.
• Demonstrated that specific water arrangements initiate autoionization and promote dissociation through local electric field fluctuations.
• Provided a detailed mechanistic insight into water autoionization, improving our understanding of its free energy landscape.