文献类型： 外文期刊

作者： Guan, Shu-hui ¹ ; Shang, Cheng ² ; Huang, Si-Da ² ; Liu, Zhi-Pan ² ;

作者机构： 1.Shanghai Acad Agr Sci, Shanghai 201403, Peoples R China

2.Fudan Univ, Collaborat Innovat Ctr Chem Energy Mat, Shanghai Key Lab Mol Catalysis & Innovat Mat, Key Lab Computat Phys Sci,Dept Chem, Shanghai 200433, Peoples R China

期刊名称：JOURNAL OF PHYSICAL CHEMISTRY C （ 影响因子：4.126； 五年影响因子：4.467 ）

ISSN： 1932-7447

年卷期： 2018 年 122 卷 50 期

页码：

收录情况： SCI

摘要： Two most known ice phases, cubic (Ic) and hexagonal ice (Ih), are observed to coexist over a surprisingly broad low temperature range (150-240 K) in ice particles, the origin of which, despite great efforts devoted to revealing the ice microstructures in experiment, remains largely unknown. Here by combining global sampling with machine learning potential, the stochastic surface walking-neural network method, we reveal all key metastable intermediates during the Ic-to-Ih phase transition, which are basically low-energy structural defects in ice, including vacancies, heterophase junctions, and stacking faults. This leads to the first identification of the lowest energy pathways from ice Ic to Ih, where the presence of vacancy defects in ice allows the transformation kinetically by initiating a fast Ih nucleation. The subsequent migration of a vacancy, with a barrier of 48 kJ/mol, turns out to be the rate-determining step in Ih phase growth, rationalizing the unexpected prevalence of anisotropic stacking faults in low-temperature ice particles. This vacancy-controlled kinetics in Ih formation arises from the flexibility of the local geometry of a weaker H-bonding environment around the vacancy, which suggests that the frustrated material with a complex energy landscape may seek an unusual phase transition pathway with the help of a low level of heterogeneity.