Predicting structure/property relationships in multi-dimensional nanoparticle data using t-distributed stochastic neighbour embedding and machine learning.

Combining researchers' domain expertise and advanced dimension reduction methods we demonstrate how visually comparing the distribution of nanoparticles mapped from multiple dimensions to a two dimensional plane can rapidly identify possible single-structure/property relationships and to a lesser extent multi-structure/property relationships. These relationships can be further investigated and confirmed with machine learning, using genetic programming to inform the choice of property-specific models and their hyper-parameters. In the case of our nanodiamond case study, we visually identify and confirm a strong relationship between the size and the probability of observation (stability) and a more complicated (and visually ambiguous) relationship between the ionisation potential and band gaps with a range of different structural, chemical and statistical surface features, making it more difficult to engineer in practice.

Nanoinformatics, and the big challenges for the science of small things.

The combination of computational chemistry and computational materials science with machine learning and artificial intelligence provides a powerful way of relating structural features of nanomaterials with functional properties. However, combining these fundamentally different scientific approaches is not as straightforward as it seems. Machine learning methods were developed for large data sets with small numbers of consistent features. Typically nanomaterials data sets are small, with high dimensionality and high variance in the feature space, and suffer from numerous destructive biases. None of the established data science or machine learning methods in widespread use today were devised with (nano)materials data sets in mind, but there are ways to overcome these challenges and use them reliably. In this review we will discuss domain-specific constraints on data-driven nanomaterials design, and explore the differences between nanomaterials simulation and nanoinformatics that can be leveraged for greater impact.

Vacancy induced formation of nanoporous silicon, carbon and silicon carbide.

Nanoporous semiconductors are used in a range of applications from sensing and gas separation, to photovoltaics, rechargeable batteries, energetic materials and micro electro mechanical systems. In most cases porosity occurs in conjunction with the competing process of amorphisation, creating a complicated material that responds differently to strain and density changes, depending on the composition. In this paper we use simple computational workflow involving Monte Carlo simulation, numerical characterisation and statistical analysis to explore the development of amorphous and nanoporous carbon, silicon and silicon carbide. We show that amorphous regions in Si and SiC form in advance of nanopores, and are essential in stabilising the nanopores once developed. Carbon prefers a porous structure at lower strains than amorphisation and exhibits a bimodal change in the structure which correlates with the change in C-C bond angles from tetrahedral sp3-like bonds to hexagonal sp2-like bonds as the strain increases. These results highlight how both of these processes can be analysed simultaneously using reliable interatomic forcefields or density functionals, provided sufficient samples are included to support the statistics.

Crystallization in suspensions of hard spheres: a Monte Carlo and molecular dynamics simulation study.

The crystallization of a metastable melt is one of the most important non-equilibrium phenomena in condensed matter physics, and hard sphere colloidal model systems have been used for several decades to investigate this process by experimental observation and computer simulation. Nevertheless, there is still an unexplained discrepancy between the simulation data and experimental nucleation rate densities. In this paper we examine the nucleation process in hard spheres using molecular dynamics and Monte Carlo simulation. We show that the crystallization process is mediated by precursors of low orientational bond-order and that our simulation data fairly match the experimental data sets.

Precursor-mediated crystallization process in suspensions of hard spheres.

We report on a large scale computer simulation study of crystal nucleation in hard spheres. Through a combined analysis of real- and reciprocal-space data, a picture of a two-step crystallization process is supported: First, dense, amorphous clusters form which then act as precursors for the nucleation of well-ordered crystallites. This kind of crystallization process has been previously observed in systems that interact via potentials that have an attractive as well as a repulsive part, most prominently in protein solutions. In this context the effect has been attributed to the presence of metastable fluid-fluid demixing. Our simulations, however, show that a purely repulsive system (that has no metastable fluid-fluid coexistence) crystallizes via the same mechanism.

Modeling of structure and porosity in amorphous silicon systems using Monte Carlo methods.

Porous solids are very important from a scientific point of view as they provide a medium in which to study the behavior of confined fluids. Although some porous solids have a well defined pore geometry such as zeolites, many porous solids lack crystalline order and are usually described as amorphous. The description of the pore geometry in such structures is very difficult. The authors develop a modeling approach using a Monte Carlo algorithm to simulate porosity within amorphous systems based on constraints for the internal volume and surface area. To illustrate this approach, a model of microporous amorphous silicon is presented. Structural aspects of the porous model are then compared against hybrid reverse Monte Carlo simulations of nonporous amorphous silicon and published results from the literature. It is found that coordination defects are predominately located at the pore surface walls.