Passivation of Mid-Gap Electronic States at Calcium Aluminosilicate Glass Surfaces upon Water Exposure: An Ab Initio Study.

When a clean glass surface is exposed to humid air, a thin water layer forms on the hydrophilic surface. Using ab initio molecular dynamics, we simulate the changes in the electronic structure of a CaO-Al2O3-SiO2 glass model upon vacuum fracture and subsequent exposure to H2O. When the glass is fractured, dangling bonds form, which lower the band gap of the surface by ∼1.8 eV compared to the bulk value due to mid-gap surface states. When H2O adsorbs onto the vacuum-fractured surface, the band gap increases to a value closer to that of the bulk band gap. Using two different hydroxylation methods, we find that the calculated band gap of the glass surface depends on the hydroxylation state. Surfaces with ∼4.5 OH/nm2 have smaller band gaps due to unfilled surface states, and surfaces with ∼2.5 OH/nm2 have larger band gaps with no apparent unfilled surface states. The resulting changes in the electronic structure, quantified by electron affinity and work function values, are hypothesized to play an important role in the electrostatic charge transfer based on the principles of surface state theory, which posit that the density of electronic surface states determines the amount of electronic charge transfer to or from material surfaces.

Prediction and Interpretation of Polymer Properties Using the Graph Convolutional Network.

We present machine learning models for the prediction of thermal and mechanical properties of polymers based on the graph convolutional network (GCN). GCN-based models provide reliable prediction performances for the glass transition temperature (T g), melting temperature (T m), density (ρ), and elastic modulus (E) with substantial dependence on the dataset, which is the best for T g (R 2 ∼ 0.9) and worst for E (R 2 ∼ 0.5). It is found that the GCN representations for polymers provide prediction performances of their properties comparable to the popular extended-connectivity circular fingerprint (ECFP) representation. Notably, the GCN combined with the neural network regression (GCN-NN) slightly outperforms the ECFP. It is investigated how the GCN captures important structural features of polymers to learn their properties. Using the dimensionality reduction, we demonstrate that the polymers are organized in the principal subspace of the GCN representation spaces with respect to the backbone rigidity. The organization in the representation space adaptively changes with the training and through the NN layers, which might facilitate a subsequent prediction of target properties based on the relationships between the structure and the property. The GCN models are found to provide an advantage to automatically extract a backbone rigidity, strongly correlated with T g, as well as a potential transferability to predict other properties associated with a backbone rigidity. Our results indicate both the capability and limitations of the GCN in learning to describe polymer systems depending on the property.

Comparison of Machine Learning Methods towards Developing Interpretable Polyamide Property Prediction.

Polyamides are often used for their superior thermal, mechanical, and chemical properties. They form a diverse set of materials that have a large variation in properties between linear to aromatic compounds, which renders the traditional quantitative structure-property relationship (QSPR) challenging. We use extended connectivity fingerprints (ECFP) and traditional QSPR fingerprints to develop machine learning models to perform high fidelity prediction of glass transition temperature (Tg), melting temperature (Tm), density (ρ), and tensile modulus (E). The non-linear model using random forest is in general found to be more accurate than linear regression; however, using feature selection or regularization, the accuracy of linear models is shown to be improved significantly to become comparable to the more complex nonlinear algorithm. We find that none of the models or fingerprints were able to accurately predict the tensile modulus E, which we hypothesize is due to heterogeneity in data and data sources, as well as inherent challenges in measuring it. Finally, QSPR models revealed that the fraction of rotatable bonds, and the rotational degree of freedom affects polyamide properties most profoundly and can be used for back of the envelope calculations for a quick estimate of the polymer attributes (glass transition temperature, melting temperature, and density). These QSPR models, although having slightly lower prediction accuracy, show the most promise for the polymer chemist seeking to develop an intuition of ways to modify the chemistry to enhance specific attributes.

Laser-induced structural modification in calcium aluminosilicate glasses using molecular dynamic simulations.

Glass structures of multicomponent oxide systems (CaO-Al2O3-SiO2) are studied using a simulated pulsed laser with molecular dynamics. The short- and intermediate-range order structures revealed a direct correlation between the transformation of Al(IV) to Al(V), regions of increased density following laser processing, inherent reduction in the average T-O-T (T = Al, Si) angle, and associated elongation of the T-O bonding distance. Variable laser pulse energies were simulated across calcium aluminosilicate glasses with high silica content (50-80%) to identify densification trends attributed to composition and laser energy. High-intensity pulsed laser effects on fictive temperature and shockwave promotion are discussed in detail for their role in glass densification. Laser-induced structural changes are found to be highly dependent on pulse energy and glass chemistry.

Constructing surface models of silicate glasses using molecular dynamics to understand the effect of pH on the hydration properties.

In this work, we use realistic silicate glass surface models, with molecular dynamics simulations, and present an algorithm for proper atomic partial charge assignment, consistent with measurable internal dipoles. The immersion energy is calculated for different silicate glass compositions in solutions of varying pH. We use molecular dynamics to elucidate the differences in the structure of water between mono- and divalent cations. The immersion energy of the glass surface is found to increase with an increase in ionic surface density and pH. This can be attributed to the stronger interaction between water and cations, as opposed to the interactions between water and silanol groups. The developed models and methods provide new insights into the structure of glass-solution interfaces and the effect of cation surface density in common nanoscale environments.

Predicting Q-Speciation in Binary Phosphate Glasses Using Statistical Mechanics.

Predicting the compositional evolution of the atomic-scale structure of oxide glasses is important for developing quantitative composition-property models. In binary phosphate glasses, the addition of network modifiers generally leads to depolymerization of the networks as described by the Q-speciation, where Q n denotes PO4 tetrahedra with n number (between 0 and 3) of bridging P-O-P linkages per tetrahedron. Upon the initial creation of nonbridging oxygens and thus partly depolymerized Q species, a variety of network former-modifier interactions exist. Here, on the basis of 31P magic angle spinning nuclear magnetic resonance spectroscopy data from the literature, we present a statistical description of the compositional evolution of Q-speciation in these glasses by accounting for the relative enthalpic and entropic contributions to the bonding preferences. We show that the entire glass structure evolution can be predicted based on experimental structural information for only a few glass compositions in each series. The model also captures the differences in bonding preferences in glasses with different field strengths (charge-to-size ratio) of the modifier cations.

Computational approaches for investigating interfacial adhesion phenomena of polyimide on silica glass.

This manuscript provides a comprehensive study of adhesion behavior and its governing mechanisms when polyimide undergoes various modes of detachment from silica glass. Within the framework of steered molecular dynamics, we develop three different adhesion measurement techniques: pulling, peeling, and sliding. Such computational methodologies can be applied to investigate heterogeneous materials with differing interfacial adhesion modes. Here, a novel hybrid potential involving a combination of the INTERFACE force field in conjunction with ReaxFF and including Coulombic and Lennard-Jones interactions is employed to study such interfaces. The studies indicate that the pulling test requires the largest force and the shortest distance to detachment as the interfacial area is separated instantaneously, while the peeling test is observed to exhibit the largest distance for detachment because it separates via line-by-line adhesion. Two kinds of polyimides, aromatic and aliphatic type, are considered to demonstrate the rigidity dependent adhesion properties. The aromatic polyimide, which is more rigid due to the stronger charge transfer complex between chains, requires a greater force but a smaller distance at detachment than the aliphatic polyimide for all of the three methodologies.

Far-from-equilibrium sheared colloidal liquids: disentangling relaxation, advection, and shear-induced diffusion.

Using high-speed confocal microscopy, we measure the particle positions in a colloidal suspension under large-amplitude oscillatory shear. Using the particle positions, we quantify the in situ anisotropy of the pair-correlation function, a measure of the Brownian stress. From these data we find two distinct types of responses as the system crosses over from equilibrium to far-from-equilibrium states. The first is a nonlinear amplitude saturation that arises from shear-induced advection, while the second is a linear frequency saturation due to competition between suspension relaxation and shear rate. In spite of their different underlying mechanisms, we show that all the data can be scaled onto a master curve that spans the equilibrium and far-from-equilibrium regimes, linking small-amplitude oscillatory to continuous shear. This observation illustrates a colloidal analog of the Cox-Merz rule and its microscopic underpinning. Brownian dynamics simulations show that interparticle interactions are sufficient for generating both experimentally observed saturations.

Structure and transport properties of polymer grafted nanoparticles.

We perform molecular dynamics simulations on a bead-spring model of pure polymer grafted nanoparticles (PGNs) and of a blend of PGNs with a polymer melt to investigate the correlation between PGN design parameters (such as particle core concentration, polymer grafting density, and polymer length) and properties, such as microstructure, particle mobility, and viscous response. Constant strain-rate simulations were carried out to calculate viscosities and a constant-stress ensemble was used to calculate yield stresses. The PGN systems are found to have less structural order, lower viscosity, and faster diffusivity with increasing length of the grafted chains for a given core concentration or grafting density. Decreasing grafting density causes depletion effects associated with the chains leading to close contacts between some particle cores. All systems were found to shear thin, with the pure PGN systems shear thinning more than the blend; also, the pure systems exhibited a clear yielding behavior that was absent in the blend. Regarding the mechanism of shear thinning at the high shear rates examined, it was found that the shear-induced decrease of Brownian stresses and increase in chain alignment, both correlate with the reduction of viscosity in the system with the latter being more dominant. A coupling between Brownian stresses and chain alignment was also observed wherein the non-equilibrium particle distribution itself promotes chain alignment in the direction of shear.