Abstract:Water freezing is ubiquitous on Earth, affecting many areas from biology to climate science and aviation technology. Probing the atomic structure in the homogeneous ice nucleation process from scratch is of great value but still experimentally unachievable. Theoretical simulations have found that ice originates from the low-mobile region with increasing abundance and persistence of tetrahedrally coordinated water molecules. However, a detailed microscopic picture of how the disordered hydrogen-bond network rearranges itself into an ordered network is still unclear. In this work, we use a deep neural network (DNN) model to "learn" the interatomic potential energy from quantum mechanical data, thereby allowing for large-scale and long molecular dynamics (MD) simulations with ab initio accuracy. The nucleation mechanism and dynamics at atomic resolution, represented by a total of 36 $\mu$s-long MD trajectories, are deeply affected by the structural and dynamical heterogeneity in supercooled water. We find that imperfectly coordinated (IC) water molecules with high mobility pave the way for hydrogen-bond network rearrangement, leading to the growth or shrinkage of the ice nucleus. The hydrogen-bond network formed by perfectly coordinated (PC) molecules stabilizes the nucleus, thus preventing it from vanishing and growing. Consequently, ice is born through competition and cooperation between IC and PC molecules. We anticipate that our picture of the microscopic mechanism of ice nucleation will provide new insights into many properties of water and other relevant materials.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the microscopic mechanism in the process of water freezing, especially under the condition of homogeneous nucleation, how the disordered hydrogen - bond network rearranges to form an ordered ice - crystal structure. Specifically, the paper focuses on the role played by incompletely coordinated (IC) water molecules in this process and how they interact with completely coordinated (PC) water molecules to jointly influence the formation, growth or shrinkage of ice nuclei. By using the deep neural network (DNN) model to conduct large - scale and long - time molecular dynamics (MD) simulations, the authors aim to reveal the atomic - scale dynamic process of ice nucleation and provide new insights into the properties of water and other related materials. The main contributions of the paper are as follows: 1. **The role of incompletely coordinated (IC) water molecules**: Research shows that IC water molecules with high mobility promote the formation and change of ice nuclei through significant hydrogen - bond network reorganization. These molecules act as "catalysts" to initiate structural changes near the ice - nucleus surface, thereby affecting the size of the ice nucleus. 2. **The stabilizing role of completely coordinated (PC) water molecules**: The hydrogen - bond network formed by PC water molecules plays a role in stabilizing the ice nucleus, preventing it from disappearing or overgrowing. This stabilizing effect is achieved by maintaining the tetrahedral coordination structure within the ice nucleus. 3. **The microscopic mechanism of ice nucleation**: The paper describes in detail the reorganization mechanism of the hydrogen - bond network during ice nucleation, including the transfer, annihilation and generation of the IC state. These mechanisms explain how disordered water gradually transforms into ordered ice. 4. **Estimation of ice nucleation rate**: Through the forward flux sampling (FFS) method, the authors accurately estimate the ice nucleation rate under homogeneous conditions and verify the validity of the classical nucleation theory (CNT). The research results are highly consistent with the experimental data, further confirming the reliability of the proposed microscopic mechanism. In conclusion, this paper not only deepens the understanding of the ice nucleation process but also provides a new perspective for other research involving phase transitions and complex behaviors.

Imperfectly coordinated water molecules pave the way for homogeneous ice nucleation

Temperature Dependence of Homogeneous Nucleation in Ice.

Bridging classical nucleation theory and molecular dynamics simulation for homogeneous ice nucleation

Homogeneous nucleation of ice

Homogeneous ice nucleation from supercooled water

Atomistic View of Homogeneous Nucleation of Water into Polymorphic Ices

Hydrogen polarity of interfacial water regulates heterogeneous ice nucleation.

Hydrogen-bond linking is crucial for growing ice VII embryos

Mechanistic Investigation of Nucleation Kinetics in Heterogeneous Ice Crystallization: the Role of Cooling Rate, Surface Energy, Surface Nanostructure, and Wetting State

Nanoclusters and Amorphous Ice Flakes Built from Water Dimers

Molecular Dynamics Simulation of Freezing Process of Water Droplets Impinging on Cold Surface

Direct Calculation of Ice Homogeneous Nucleation Rate for a Molecular Model of Water

Different intermediate water cluster with distinct nucleation dynamics among mono layer ice nucleation

The Microscopic Features of Heterogeneous Ice Nucleation May Affect the Macroscopic Morphology of Atmospheric Ice Crystals

The Many Faces of Heterogeneous Ice Nucleation: Interplay Between Surface Morphology and Hydrophobicity

Theoretical prediction of the homogeneous ice nucleation rate: disentangling thermodynamics and kinetics

New metastable form of ice and its role in the homogeneous crystallization of water

A Generalized Nucleation Theory for Ice Crystallization

Homogeneous nucleation in supercooled liquid water. Determination of ice germ size and activation energy barrier in Molecular Dynamics simulations

Using machine learning with atomistic surface and local water features to predict heterogeneous ice nucleation

Role of Nanoscale Interfacial Proximity in Contact Freezing in Water