We use large-scale simulations to investigate the dynamic fracture of silica and sodium-silicate glasses under uniaxial tension. The stress-strain curves demonstrate that silica glass is brittle whereas the glasses rich in Na show pronounced ductility. A strong composition dependence is also seen in the crack velocity which is on the order of 1800 m/s for glasses with low Na concentration and decreases to 700 m/s if the concentration is high. We find that during the fracture of Na-rich glasses very irregular cavities as large as 3-4 nm form ahead of the crack front, indicating the presence of nanoductility in these glasses. Before fracture occurs, the local composition, structure, and mechanical properties are heterogeneous in space and show a strong dependence on the applied strain. Further analysis of the correlations between these local properties allows to obtain a better microscopic understanding of the deformation and fracture of glasses and how the local heating close to the crack tip, up to several hundred degrees, permits the structure to relax.
Structure and properties of magnesium silicate and borate melts and glasses were investigated by using newly parameterized interaction potentials in molecular dynamics simulations and compared with those of calcium silicate and borate. The competition between the depolymerization of the silica network and the formation of new bonds with modifier ions leads to the enhancement of the elastic moduli with increasing modifier content in alkaline earth silicate glasses. Compared with calcium silicate, the higher elastic moduli of magnesium silicate result from a higher connectivity of the overall gl…
Developing accurate, transferable, and computationally-efficient interatomic forcefields is key to facilitate the modeling of silicate glasses. However, the high number of forcefield parameters that need to be optimized render traditional parameterization methods poorly efficient or potentially subject to bias. Here, we present a new forcefield parameterization methodology based on ab initio molecular dynamics simulations, Gaussian process regression, and Bayesian optimization. By taking the example of glassy silica, we show that our methodology yields a new interatomic forcefield that offers …
Predicting the dissolution rates of silicate glasses in aqueous conditions is a complex task as the underlying mechanism(s) remain poorly understood and the dissolution kinetics can depend on a large number of intrinsic and extrinsic factors. Here, we assess the potential of data-driven models based on machine learning to predict the dissolution rates of various aluminosilicate glasses exposed to a wide range of solution pH values, from acidic to caustic conditions. Four classes of machine learning methods are investigated, namely, linear regression, support vector machine regression, random f…
The development by machine learning of models predicting materials' properties usually requires the use of a large number of consistent data for training. However, quality experimental datasets are not always available or self-consistent. Here, as an alternative route, we combine machine learning with high-throughput molecular dynamics simulations to predict the Young's modulus of silicate glasses. We demonstrate that this combined approach offers excellent predictions over the entire compositional domain. By comparing the performance of select machine learning algorithms, we discuss the natur…
We use \textit{ab initio} molecular dynamics simulations to investigate the properties of the dry surface of pure silica and sodium silicate glasses. The surface layers are defined based on the atomic distributions along the direction ($z-$direction) perpendicular to the surfaces. We show that these surfaces have a higher concentration of dangling bonds as well as two-membered (2M) rings than the bulk samples. Increasing concentration of Na$_2$O reduces the proportion of structural defects. From the vibrational density of states, one concludes that 2M rings have a unique vibrational signature …
For small tension the response of a solid to an applied stress is given by Hooke's law. Outside this linear regime the relation between stress and strain is no longer universal and at present there is no satisfactory insight on how to connect for disordered materials the stress-strain relation to the microscopic properties of the system. Here we use atomistic computer simulations to establish this connection for the case of silicate glasses containing modifiers. By probing how in the highly non-linear regime the stress-strain curve depends on composition, we are able to identify the microscopi…
Density is one of the most commonly measured or estimated materials properties, especially for glasses and melts that are of significant interest to many fields, including metallurgy, geology, materials science and sustainable cements. Here, two types of machine learning (ML) models (i.e., random forest (RF) and artificial neural network (ANN)) have been developed to predict the room-temperature density of glasses in the compositional space of CaO-MgO-Al2O3-SiO2-TiO2-FeO-Fe2O3-Na2O-K2O-MnO (CMASTFNKM), based on ~2100 data points mined from ~140 literature studies. The results show that the RF …
Using molecular dynamics simulations we investigate the dependence of the structural and vibrational properties of the surfaces of sodo-silicate glasses on the sodium content as well as the nature of the surface. Two types of glass surfaces are considered: A melt-formed surface (MS) in which a liquid with a free surface has been cooled down into the glass phase and a fracture surface (FS) obtained by tensile loading of a glass sample. We find that the MS is more abundant in Na and non-bridging oxygen atoms than the FS and the bulk glass, whereas the FS has higher concentration of structural de…
We compare the ability of various interaction potentials to predict the structural and mechanical properties of silica and sodium silicate glasses. While most structural quantities show a relatively mild dependence on the potential used, the mechanical properties such as the failure stress and strain as well as the elastic moduli depend very strongly on the potential, once finite size effects have been taken into account. We find that to avoid such finite size effects, samples of at least 75,000 atoms are needed. Finally we probe how the simulation ensemble influences the fracture properties o…
We apply a recently developed optimization scheme to obtain effective potentials for alkali and alkaline-earth aluminosilicate glasses that contains lithium, sodium, potassium, or calcium as modifiers. As input data for the optimization, we used the radial distribution functions of the liquid at high temperature generated by means of ab initio molecular dynamics simulations and density and elastic modulus of glass at room temperature from experiments. The new interaction potentials are able to reproduce reliably the structure and various mechanical and vibrational properties over a wide range …
Topological constraint theory has become an increasingly popular tool to predict the compositional dependence of glass properties or pinpoint promising compositions with tailored functionalities. This approach reduces complex disordered networks into simpler mechanical trusses. Thereby, topological constraint theory captures the important atomic topology that controls macroscopic properties while filtering out less relevant second-order structural details. As such, topological constraint theory can be used to decode the genome of glass, that is, to identify and decipher how the basic structura…
The Gardner length scale $ξ$ is the correlation length in the vicinity of the Gardner transition, which is an avoided transition in glasses where the phase space of the glassy phase fractures into smaller sub-basins on experimental time scales. We argue that $ξ$ grows like $ \sim \sqrt{B_{\infty}/G_{\infty}}$, where $B_{\infty}$ is the bulk modulus and $G_{\infty}$ is the shear modulus, both measured in the high-frequency limit of the glassy state. We suggest that $ξ$ might be inferred from stress-stress correlation functions, which is more practical for experimental investigation than studyin…
The deformation of disordered solids relies on swift and localised rearrangements of particles. The inspection of soft vibrational modes can help predict the locations of these rearrangements, while the strain that they actually redistribute mediates collective effects. Here, we study soft modes and strain redistribution in a two-dimensional continuous mesoscopic model based on a Ginzburg-Landau free energy for perfect solids, supplemented with a plastic disorder potential that accounts for shear softening and rearrangements. Regardless of the disorder strength, our numerical simulations show …
