Periodic Reporting for period 4 - SOFTWATER (Soft Water: understanding what makes a fluid behave like water)
Reporting period: 2021-05-01 to 2022-10-31
Water is unique in many ways. It can freeze in multiple ice forms (polymorphism) and can vitrify in multiple glass forms (polyamorphism). The freezing of water is arguably the most important phase transition on Earth, playing a fundamental role in the energy balance of our planet (e.g. cloud albedo). The vitrification of water, instead, produces the most abundant form of water in the universe, found for example in the core of giant icy planets. Despite the importance of these solid states, we still lack a microscopic understanding of the transition from liquid water to a solid phase.
One of the biggest challenges in these types of studies is to develop mathematical quantities that allow us to characterize the local environment around water molecules, and the changes that occur at extreme conditions.
Our project goal is to develop new methods of local structure characterization, based on high-dimensional and unbiased order parameters, and to use them to formulate a unified description of all water anomalies. We hope that our approach will open new possibilities in the study of water at extreme conditions, transforming the way nucleation and vitrification are analyzed, and offer new microscopic insights into these processes.
One of the biggest challenges in these types of studies is to develop mathematical quantities that allow us to characterize the local environment around water molecules, and the changes that occur at extreme conditions.
Our project goal is to develop new methods of local structure characterization, based on high-dimensional and unbiased order parameters, and to use them to formulate a unified description of all water anomalies. We hope that our approach will open new possibilities in the study of water at extreme conditions, transforming the way nucleation and vitrification are analyzed, and offer new microscopic insights into these processes.
The goal of the research that I am undertaking within the current project is to understand the behaviour of water from the perspective of coarse-grained Soft Matter models. During the first 18 months, I've pursued this topic following the research directions that were identified as working packages (WP) in the original proposal.
WP1) The first work package’s goal is to develop new theoretical and computational methods to study water anomalies. Progress has been made in two different areas.
1) tuning water’s potential to interpolate continuously between a simple liquid and a tetrahedrally coordinated liquid. In coarse-grained representations of water, we are able to tune the tetrahedrality and study how the properties of water change. To characterize these changes to the structure, we decided to use new topological measures, and in particular persistent homology.
Our preliminary work has been published in, for which I partnered with experts in topological descriptions of materials:
* A new topological descriptor for water network structure, L. Steinberg, J. Russo, and J. Frey, J. of Cheminformatics 11, 48 (2019)
We shown that differences in water models are able to be isolated to different degrees of homology.
2) detecting local structures in water. We have developed a novel order parameter which leverages the power of neural networks to identify both crystalline polymorphs and polyamorphic amorphous phases in water.
WP2) The second work package aims at understanding the anomalous behavior of the liquid phase. We have made considerably progress on two fronts.
1) understanding thermodynamic anomalies. The following publication details our progress
*) Water-like anomalies as a function of tetrahedrality, J. Russo, K. Akahane, and H. Tanaka, PNAS 115, E3333, 9 pages (2018)
Here we’ve completed a study of how the free energy landscape of water changes as a function of its tetrahedrality. We show that by tuning tetrahedral interactions, we can continuously interpolate between the behaviour predicted by different water scenarios. We also rationalize the behaviour observed in computer simulations with a two-state water model, that provide both a quantitative and qualitative understanding of all the anomalies.
2) understanding dynamic anomalies. Currently the dynamic behaviour of water at low temperatures is ascribed to glassy phenomenology. Our work aims at radically changing our understanding of the dynamic anomalies (including its pressure dependence) without resorting to any singularity (either dynamic or thermodynamic). A new hierarchical two-state model is first introduced in the following two publications
*) Common microscopic structural origin for water’s thermodynamic and dynamic anomalies, R. Shi, J. Russo, and H. Tanaka, J. Chem. Phys. 149, 224502 (2019)
WP3) This work package aims at understanding the behavior of glassy water.
*) Glass forming ability in systems with competing orderings, J. Russo, F. Romano, and H. Tanaka, Physical Review X, 8, 021040, 17 pages (2018)
In this manuscript we have demonstrated that glass-forming ability is correlated with a single thermodynamic quantity, that we call thermodynamic interface penalty. Aside from water, our framework is applicable to all systems with competing interactions, e.g. metallic glass formers.
WP4) This work package aims at understanding the process of ice nucleation. The PDRA of the project (Dr. Fabio Leoni) has been working extensively on this work package, and a paper has been published, and another one is under submission
*) Crystalline clusters in mW water: Stability, growth, and grain boundaries, F. Leoni, R. Shi, H. Tanaka, and J. Russo, J. Chem. Phys. 151, 044505 (2019)
Here we aim at understanding nucleation rates in water. We employ Soft Matter models in order to understand the process of nucleation in supercooled water. We systematically investigate all different contributions to nucleation, and in particular the role of surface free energy.
WP1) The first work package’s goal is to develop new theoretical and computational methods to study water anomalies. Progress has been made in two different areas.
1) tuning water’s potential to interpolate continuously between a simple liquid and a tetrahedrally coordinated liquid. In coarse-grained representations of water, we are able to tune the tetrahedrality and study how the properties of water change. To characterize these changes to the structure, we decided to use new topological measures, and in particular persistent homology.
Our preliminary work has been published in, for which I partnered with experts in topological descriptions of materials:
* A new topological descriptor for water network structure, L. Steinberg, J. Russo, and J. Frey, J. of Cheminformatics 11, 48 (2019)
We shown that differences in water models are able to be isolated to different degrees of homology.
2) detecting local structures in water. We have developed a novel order parameter which leverages the power of neural networks to identify both crystalline polymorphs and polyamorphic amorphous phases in water.
WP2) The second work package aims at understanding the anomalous behavior of the liquid phase. We have made considerably progress on two fronts.
1) understanding thermodynamic anomalies. The following publication details our progress
*) Water-like anomalies as a function of tetrahedrality, J. Russo, K. Akahane, and H. Tanaka, PNAS 115, E3333, 9 pages (2018)
Here we’ve completed a study of how the free energy landscape of water changes as a function of its tetrahedrality. We show that by tuning tetrahedral interactions, we can continuously interpolate between the behaviour predicted by different water scenarios. We also rationalize the behaviour observed in computer simulations with a two-state water model, that provide both a quantitative and qualitative understanding of all the anomalies.
2) understanding dynamic anomalies. Currently the dynamic behaviour of water at low temperatures is ascribed to glassy phenomenology. Our work aims at radically changing our understanding of the dynamic anomalies (including its pressure dependence) without resorting to any singularity (either dynamic or thermodynamic). A new hierarchical two-state model is first introduced in the following two publications
*) Common microscopic structural origin for water’s thermodynamic and dynamic anomalies, R. Shi, J. Russo, and H. Tanaka, J. Chem. Phys. 149, 224502 (2019)
WP3) This work package aims at understanding the behavior of glassy water.
*) Glass forming ability in systems with competing orderings, J. Russo, F. Romano, and H. Tanaka, Physical Review X, 8, 021040, 17 pages (2018)
In this manuscript we have demonstrated that glass-forming ability is correlated with a single thermodynamic quantity, that we call thermodynamic interface penalty. Aside from water, our framework is applicable to all systems with competing interactions, e.g. metallic glass formers.
WP4) This work package aims at understanding the process of ice nucleation. The PDRA of the project (Dr. Fabio Leoni) has been working extensively on this work package, and a paper has been published, and another one is under submission
*) Crystalline clusters in mW water: Stability, growth, and grain boundaries, F. Leoni, R. Shi, H. Tanaka, and J. Russo, J. Chem. Phys. 151, 044505 (2019)
Here we aim at understanding nucleation rates in water. We employ Soft Matter models in order to understand the process of nucleation in supercooled water. We systematically investigate all different contributions to nucleation, and in particular the role of surface free energy.
For the different work packages described above, the progress beyond the start of the art and expected results are.
WP1) The goal of this work package is to effectively develop our new order parameters and the corresponding neural networks. In particular, our order parameter will be extended to include information about molecular orientation, e.g. adding hydrogens. The Neural Network will also be optimized before applications to the other work packages. We will explore Machine Learning techniques, such as dimensionality reduction and metric learning, to effectively reduce the computational effort, while ensuring a good partitioning of the state space.
WP3) In this work package we address disordered local configurations, i.e. amorphous local structures. Water is characterized by having a vast array of amorphous glasses, for example LDA, HDA, ASW, to name just a few. This provides the unique opportunity to compare these different amorphous configurations with our new local measure. The main goals of this WP are the following
* Detect structural differences between amorphous ices
* Explore the connection between structural glasses and the supercooled liquid state.
* Investigate the nature of the glass transitions in terms of molecular translations and rotations
WP4) In this work package we use the methodologies developed in WP1 to study the process of nucleation in water at supercooled conditions. The main goals of this WP are:
* Understand the polymorph selection problem in water. While most ice is found in the hexagonal phase, during nucleation it is known that stacking disordered ice is formed instead.
* Computing nucleation barriers and nucleation rates. Combined with rare-event simulation techniques (such as Umbrella Sampling) our new methodology will allow to characterize the behaviour of water at conditions relevant to climate models.
* Investigate the role of metastable phases. Several tetrahedral crystalline clusters exist in water, often with five-membered rings, which are energetically favoured at small scales.
* Provide a general order parameter to be used for different potentials. Comparison with experimental data will be performed.
WP1) The goal of this work package is to effectively develop our new order parameters and the corresponding neural networks. In particular, our order parameter will be extended to include information about molecular orientation, e.g. adding hydrogens. The Neural Network will also be optimized before applications to the other work packages. We will explore Machine Learning techniques, such as dimensionality reduction and metric learning, to effectively reduce the computational effort, while ensuring a good partitioning of the state space.
WP3) In this work package we address disordered local configurations, i.e. amorphous local structures. Water is characterized by having a vast array of amorphous glasses, for example LDA, HDA, ASW, to name just a few. This provides the unique opportunity to compare these different amorphous configurations with our new local measure. The main goals of this WP are the following
* Detect structural differences between amorphous ices
* Explore the connection between structural glasses and the supercooled liquid state.
* Investigate the nature of the glass transitions in terms of molecular translations and rotations
WP4) In this work package we use the methodologies developed in WP1 to study the process of nucleation in water at supercooled conditions. The main goals of this WP are:
* Understand the polymorph selection problem in water. While most ice is found in the hexagonal phase, during nucleation it is known that stacking disordered ice is formed instead.
* Computing nucleation barriers and nucleation rates. Combined with rare-event simulation techniques (such as Umbrella Sampling) our new methodology will allow to characterize the behaviour of water at conditions relevant to climate models.
* Investigate the role of metastable phases. Several tetrahedral crystalline clusters exist in water, often with five-membered rings, which are energetically favoured at small scales.
* Provide a general order parameter to be used for different potentials. Comparison with experimental data will be performed.