اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدExtracting ice phases from liquid water: why a machine-learning water model generalizes so well
131
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
We investigate the structural similarities between liquid water and 53 ices, including 20 knowncrystalline phases. We base such similarity comparison on the local environments that consist of atoms within a certain cutoff radius of a central atom. We reveal that liquid water explores the localenvironments of the diverse ice phases, by directly comparing the environments in these phases using general atomic descriptors, and also by demonstrating that a machine-learning potential trained on liquid water alone can predict the densities, the lattice energies, and vibrational properties of theices. The finding that the local environments characterising the different ice phases are found in water sheds light on water phase behaviors, and rationalizes the transferability of water models between different phases.
قيم البحث
اقرأ أيضاً
The segmental specific heat ratio of the couple hydrogen bond defines not only the phase of Vapor, Liquid, Ice I and XI phase with a quasisolid phase that shows the negative thermal extensibility but uniquely the slope of density of water ice in diff
erent phases. Ice floats because H-O contracts less than O:H expands in the QS phase at cooling.
We introduce a coarse-grained deep neural network model (CG-DNN) for liquid water that utilizes 50 rotational and translational invariant coordinates, and is trained exclusively against energies of ~30,000 bulk water configurations. Our CG-DNN potent
ial accurately predicts both the energies and molecular forces of water; within 0.9 meV/molecule and 54 meV/angstrom of a reference (coarse-grained bond-order potential) model. The CG-DNN water model also provides good prediction of several structural, thermodynamic, and temperature dependent properties of liquid water, with values close to that obtained from the reference model. More importantly, CG-DNN captures the well-known density anomaly of liquid water observed in experiments. Our work lays the groundwork for a scheme where existing empirical water models can be utilized to develop fully flexible neural network framework that can subsequently be trained against sparse data from high-fidelity albeit expensive beyond-DFT calculations.
The dielectric spectrum of liquid water, $10^{4} - 10^{11}$ Hz, is interpreted in terms of diffusion of charges, formed as a result of self-ionization of H$_{2}$O molecules. This approach explains the Debye relaxation and the dc conductivity as two m
anifestations of this diffusion. The Debye relaxation is due to the charge diffusion with a fast recombination rate, $1/tau_{2}$, while the dc conductivity is a manifestation of the diffusion with a much slower recombination rate, $1/tau_{1}$. Applying a simple model based on Brownian-like diffusion, we find $tau_{2} simeq 10^{-11}$ s and $tau_{1} simeq 10^{-6}$ s, and the concentrations of the charge carriers, involved in each of the two processes, $N_{2} simeq 5 times 10^{26}$ m$^{-3}$ and $N_{1} simeq 10^{14}$ m$^{-3}$. Further, we relate $N_{2}$ and $N_{1}$ to the total concentration of H$_{3}$O$^{+}$--OH$^{-}$ pairs and to the pH index, respectively, and find the lifetime of a single water molecule, $tau_{0} simeq 10^{-9}$ s. Finally, we show that the high permittivity of water results mostly from flickering of separated charges, rather than from reorientations of intact molecular dipoles.
Sound driven gas bubbles in water can emit light pulses (sonoluminescence). Experiments show a strong dependence on the type of gas dissolved in water. Air is found to be one of the most friendly gases towards this phenomenon. Recently, cite{loh96} h
ave suggested a chemical mechanism to account for the strong dependence on the gas mixture: the dissociation of nitrogen at high temperatures and its subsequent chemical reactions to highly water soluble gases such as NO, NO$_2$, and/or NH$_3$. Here, we analyze the consequences of the theory and offer detailed comparison with the experimental data of Puttermans UCLA group. We can quantitatively account for heretofore unexplained results. In particular, we understand why the argon percentage in air is so essential for the observation of stable SL.
Aptly named, ice giants such as Uranus and Neptune contain significant amounts of water. While this water cannot be present near the cloud tops, it must be abundant in the deep interior. We investigate the likelihood of a liquid water ocean existing
in the hydrogen-rich region between the cloud tops and deep interior. Starting from an assumed temperature at a given upper tropospheric pressure (the photosphere), we follow a moist adiabat downward. The mixing ratio of water to hydrogen in the gas phase is small in the photosphere and increases with depth. The mixing ratio in the condensed phase is near unity in the photosphere and decreases with depth; this gives two possible outcomes. If at some pressure level the mixing ratio of water in the gas phase is equal to that in the deep interior, then that level is the cloud base. Alternately, if the mixing ratio of water in the condensed phase reaches that in the deep interior, then the surface of a liquid ocean will occur. We find that Neptune is both too warm (photospheric temperature too high) and too dry (mixing ratio of water in the deep interior too low) for liquid oceans to exist at present. To have a liquid ocean, Neptunes deep interior water to gas ratio would have to be higher than current models allow, and the density at 19 kbar would have to be ~ 0.8 g/cm^3. Such a high density is inconsistent with gravitational data obtained during the Voyager flyby. As Neptune cools, the probability of a liquid ocean increases. Extrasolar hot Neptunes, which presumably migrate inward toward their parent stars, cannot harbor liquid water oceans unless they have lost almost all of the hydrogen and helium from their deep interiors.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
