Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userTopological Descriptors Help Predict Guest Adsorption in Nanoporous Materials
148
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
Machine learning has emerged as an attractive alternative to experiments and simulations for predicting material properties. Usually, such an approach relies on specific domain knowledge for feature design: each learning target requires careful selection of features that an expert recognizes as important for the specific task. The major drawback of this approach is that computation of only a few structural features has been implemented so far, and it is difficult to tell a priori which features are important for a particular application. The latter problem has been empirically observed for predictors of guest uptake in nanoporous materials: local and global porosity features become dominant descriptors at low and high pressures, respectively. We investigate a feature representation of materials using tools from topological data analysis. Specifically, we use persistent homology to describe the geometry of nanoporous materials at various scales. We combine our topological descriptor with traditional structural features and investigate the relative importance of each to the prediction tasks. We demonstrate an application of this feature representation by predicting methane adsorption in zeolites, for pressures in the range of 1-200 bar. Our results not only show a considerable improvement compared to the baseline, but they also highlight that topological features capture information complementary to the structural features: this is especially important for the adsorption at low pressure, a task particularly difficult for the traditional features. Furthermore, by investigation of the importance of individual topological features in the adsorption model, we are able to pinpoint the location of the pores that correlate best to adsorption at different pressure, contributing to our atom-level understanding of structure-property relationships.
rate research
Read More
Nanoporous materials are a promising platform for thermoelectrics in that they offer high thermal conductivity tunability while preserving good electrical properties, a crucial requirement for high- effciency thermal energy conversion. Understanding the impact of the pore arrangement on thermal transport is pivotal to engineering realistic materials, where pore disorder is unavoidable. Although there has been considerable progress in modeling thermal size effects in nanostructures, it has remained a challenge to screen such materials over a large phase space due to the slow simulation time required for accurate results. We use density functional theory in connection with the Boltzmann transport equation, to perform calculations of thermal conductivity in disordered porous materials. By leveraging graph theory and regressive analysis, we identify the set of pores representing the phonon bottleneck and obtain a descriptor for thermal transport, based on the sum of the pore-pore distances between such pores. This approach provides a simple tool to estimate phonon suppression in realistic porous materials for thermoelectric applications and enhance our understanding of heat transport in disordered materials.
Magnetic coercivity is often viewed to be lower in alloys with negligible (or zero) values of the anisotropy constant. However, this explains little about the dramatic drop in coercivity in FeNi alloys at a non-zero anisotropy value. Here, we develop a theoretical and computational tool to investigate the fundamental interplay between material constants that govern coercivity in bulk magnetic alloys. The two distinguishing features of our coercivity tool are that: (a) we introduce a large localized disturbance, such as a spike-like magnetic domain, that provides a nucleation barrier for magnetization reversal; and (b) we account for magneto-elastic energy -- however small -- in addition to the anisotropy and magnetostatic energy terms. We apply this coercivity tool to show that the interactions between local instabilities and material constants, such as anisotropy and magnetostriction constants, are key factors that govern magnetic coercivity in bulk alloys. Using our model, we show that coercivity is minimum at the permalloy composition (Fe-21.5Ni-78.5) at which the alloys anisotropy constant is not zero. We systematically vary the values of the anisotropy and magnetostriction constants, around the permalloy composition, and identify new combinations of material constants at which coercivity is small. More broadly, our coercivity tool provides a theoretical framework to potentially discover novel magnetic materials with low coercivity.
The ground state of $^4$He confined in a system with the topology of a cylinder can display properties of a solid, superfluid and liquid crystal. This phase, which we call compactified supersolid (CSS), originates from wrapping the basal planes of the bulk hcp solid into concentric cylindrical shells, with several central shells exhibiting superfluidity along the axial direction. Its main feature is the presence of a topological defect which can be viewed as a disclination with Frank index $n=1$ observed in liquid crystals, and which, in addition, has a superfluid core. The CSS as well as its transition to an insulating compactified solid with a very wide hysteresis loop are found by ab initio Monte Carlo simulations. A simple analytical model captures qualitatively correctly the main property of the CSS -- a gradual decrease of the superfluid response with increasing pressure.
Porous carbonaceous materials have many important industrial applications including energy storage, water purification, and adsorption of volatile organic compounds. Most of their applications rely upon the adsorption of molecules or ions within the interior pore volume of the carbon particles. Understanding the behaviour and properties of adsorbate species on the molecular level is therefore key for optimising porous carbon materials, but this is very challenging owing to the complexity of the disordered carbon structure and the presence of multiple phases in the system. In recent years, NMR spectroscopy has emerged as one of the few experimental techniques that can resolve adsorbed species from those outside the pore network. Adsorbed, or in-pore species give rise to resonances that appear at lower chemical shifts compared to their free (or ex-pore) counterparts. This shielding effect arises primarily due to ring currents in the carbon structure in the presence of a magnetic field, such that the observed chemical shift differences upon adsorption are nucleus-independent to a first approximation. Theoretical modelling has played an important role in rationalising and explaining these experimental observations. Together, experiments and simulations have enabled a large amount of information to be gained on the adsorption and diffusion of adsorbed species, as well as on the structural and magnetic properties of the porous carbon adsorbent. Here, we review the methodological developments and applications of NMR spectroscopy and related modelling in this field, and provide perspectives on possible future applications and research directions.
Geometric information such as the space groups and crystal systems plays an important role in the properties of crystal materials. Prediction of crystal system and space group thus has wide applications in crystal material property estimation and structure prediction. Previous works on experimental X-ray diffraction (XRD) and density functional theory (DFT) based structure determination methods achieved outstanding performance, but they are not applicable for large-scale screening of materials compositions. There are also machine learning models using Magpie descriptors for composition based material space group determination, but their prediction accuracy only ranges between 0.638 and 0.907 in different kinds of crystals. Herein, we report an improved machine learning model for predicting the crystal system and space group of materials using only the formula information. Benchmark study on a dataset downloaded from Materials Project Database shows that our random forest models based on our new descriptor set, achieve significant performance improvements compared with previous work with accuracy scores ranging between 0.712 and 0.961 in terms of space group classification. Our model also shows large performance improvement for crystal system prediction. Trained models and source code are freely available at url{https://github.com/Yuxinya/SG_predict}
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
