Session E56: Computational Design and Discovery of Novel Materials

Live

The work function of organic/metallic interfaces can be readily tuned via the interface dipole that emerges when adsorbing organic electron-donor or -acceptor molecules. Despite extensive research in the field and the wealth of experimental data available, it is presently still not possible to predict the magnitude of the dipole based on the choice of the interface materials alone.
In this contribution we explain how data obtained through high throughput calculations and experiments can be utilized to find an explicit formula for important interface properties such as the interface dipole. This is done via the Sure Independence Screening Sparsifying Operator (SISSO), which uses compressed sensing and dimensionality reduction to find simple yet accurate regression models for the properties of the isolated components and the combined interface. We demonstrate how these models can be analyzed to extract physical insight and how to avoid potential pitfalls.   

Live

Blending a semiconducting polymer with an insulator can significantly increase [1] luminous efficiency of polymer light emitting diodes. However, the limited thermodynamic stability of the disordered phase in blends motivates the search for alternatives, e.g. block-copolymers (BCPs) with semiconducting and insulating blocks. We choose as model systems blends and BCPs of poly(p-phenylene vinylene) (PPV) and polyacrylates. We obtain disordered morphologies of these materials using mesoscopic simulations. We study different compositions and vary immiscibility to mimic annealing at different temperatures. We find that disordered blends and BCPs are heterogeneous because of fluctuations and local segregation. Local segregation is stronger in BCPs than in their equivalent blends, even though the strength of immiscibility, normalized by the mean-field spinodal, is the same. Using a qualitative charge-percolation model, we link the distribution of PPV with electric conductance. We predict [2] that the annealing temperature affects the electrical percolation in disordered BCPs much stronger than in blends. The differences between blends and BCPs are enhanced at high contents of insulator. [1] Abbaszadeh et al, Nature Materials 2016, 15, 628; [2] Zhang et al, Macromolecules 2020, 53, 523   

Live

Research data paired with Artificial Intelligence (AI) enable of a new level, a new quality of science. The ultimate goal in our research domain is to predict novel candidate materials for a given application, possibly even in regions of the materials space that no-one would think of. A real breakthrough is, however, only possible if a few key prerequisites are brought together: Big Data – meaning also the relevant data and reliable data – and novel AI tools with predictive power, all combined in a FAIR data sharing platform. In 2014, the Novel Materials Discovery (NOMAD) Laboratory (https://nomad-lab.eu) set out to make this happen for computational materials science. For reaching the ultimate goal, data from synthesis, experiment, and theory must be brought together. I’ll discuss where we are on this road.   

Live

The electronic density of states (DOS) describes the energy levels accessible to electrons in a quasi-particle picture. It is essential to interpret experimental observables such as heat capacity, or optical absorption. In this work, we present a machine-learning (ML) model to learn the DOS in different classes of materials, discuss the challenges of predicting it, provide insights on more complex systems and quantify finite-temperature effects. We introduce an atom-centered model for the electronic DOS where we expand the total DOS of a structure into a sum of contributions from its atomic environments. This ML model provides quantitative predictions of the DOS and its derived quantities like the band energy in our validation sets. We successfully apply the model to predict a hybrid-DFT quality DOS of large amorphous Silicon structures covering a wide range of pressures. We also employ the model to account for the electronic contributions to the heat capacity in metallic systems. This approach demonstrates the impact of a universal model describing structural and electronic properties inexpensively and its ability to enable more accurate and predictive materials modeling and design.   

Live

Topological materials have been attracting considerable attention due to their exotic physical properties. Utilizing first-principles methods, we predict a new topological candidate of structure type in the tetragonal p4/nmm space group (no.129) closely match with the existing ZrSiS ternary compound. The electronic band structure displays several Dirac crossings with the coexistence of both type I and type II Dirac points. Remarkably, type I Dirac crossings are robust against the spin-orbital coupling, and type II Dirac crossings are protected by the spin-orbital coupling. Formation energy calculation through the density functional theory (DFT) allows us to predict the stability of the structure. Elastic constants calculations verify the Born mechanical stability criteria of the compound. The compound is suggested for further studies for Dirac electrons.   

Live

Organic mixed ion-electron conducting (OMIEC) polymers exhibit transport of both electrons and ions. This unique functionality underpins many emerging applications such as, biosensors, organic electrochemical transistors, neurocomputing devices, and batteries. OMIEC design is incipient, and the few materials that have been synthesized have been based on electron conducting units and ion conducting units from adjacent applications in organic semiconductors and polymer electrolytes, respectively. Although this modular approach has produced materials exhibiting mixed transport, further progress is frustrated by incomplete knowledge of how ionic and electronic transport are coupled in these materials and the absence of design rules specific to OMIECs. In this work, we develop general coarse-grained model which can effectively explore the slow dynamics and nanoscale features of these systems. The model is extremely flexible and enables variation of backbone anisotropy, persistence length, side-chain density, hydrophilicity, and patterning, that can be used to interrogate how these general properties affect OMIEC behavior and electronic ionic coupling. We present our findings on the emergent OMIEC physics upon variation in sidechain composition and further, suggest novel design rules.   

Live

Precious metal nanoparticles deposited oxide supports are known to be good catalysts for several commercially important chemical reactions. The catalytic activity of such nanocatalysts is known to be highly sensitive to their morphology and charge. It is known that differently charged metal nanocatalysts are to be preferred for different chemical reactions; it is also generally believed that two-dimensional planar nanoparticles would be better catalysts than three-dimensional ones. Among the various ways that have been suggested to gain control over the nanoparticle morphology and charge, possibly one of the simplest is to dope the oxide support with an electron donor or acceptor. Using density functional theory calculations, we have shown that doping with an electron donor can switch the morphology of small deposited gold nanoparticles from upright or clumped-up shapes to wetting geometries, due to a charge transfer from the dopant atoms to the nanocatalyst. We have identified two possible descriptors for the efficacy of doping; one of these depends on the electronegativities and atomic sizes of the dopant atoms and the cation/anion in the host support. Another descriptor, that correlates well with both the morphology and charge of the nanocatalyst, is the work function of the bare support. Depending upon the type of dopant, one can make the deposited Au nanoparticles positively or negatively charged. We have also investigated the effect of support doping on adsorption energies and activation barriers for simple reactions.   

Live

The feasibility and the stability of a defect, in the host lattice is usually obtained via experiments and/or through detailed quantum mechanical calculations. Both of these conventional routes are expensive and time consuming. An alternative is a data-driven machine learning (ML)-based approach. Here, using ML techniques we identify the factors that influence defect formation energy in two material classes namely perovskites and MXenes. Using elemental properties as features and random forest regression, we demonstrate a systematic approach to down select the important features, establishing a framework for accurate predictions of the defect formation energy. Our work reveals previously unknown correlations, chemical trends, and the interplay between stability and underlying chemistries. Hence, these results showcase the efficacy of ML tools in identifying and quantifying different feature-dependencies and provide a promising route toward dopant selection. The framework itself is general and can be applied to other material classes.   

Live

Engineering micro/nanoscale structures through self-assembly of colloidal building blocks is a powerful approach in material design and processing. However, the complexity of these building blocks makes it hard to predict the resulting self-assembled structures and their transport properties. Therefore, developing predictive tools based on computational and machine learning approaches becomes a necessity for successful implementation of new material design procedures.

In this study, we establish a Manifold Learning-based approach to investigate the self-assembly process of colloidal building blocks by transforming the high-dimensional energy landscapes of colloidal particles into low-dimensional parameter spaces. This illustration of the phase space, termed the ξ space, compresses the full energy profile of the system into a single snapshot that contains important information regarding the assembly process and allows the comparison of the behavior of constructs, with different geometric features, in the same abstract space. This approach also facilitates the discovery of intuitive self-assembly pathways based on the energy footprint of colloidal particles, and may pave the way for developing new tools for material discovery and identification.   

Live

Mixed-anion materials are interesting counterparts to their more widely studied single-anion compounds due to the increased flexibility of properties afforded by the presence of multiple anions. Here, we demonstrate how computational approaches, based on DFT datasets can be combined with materials informatics and machine learning (ML) models to accelerate materials discovery. We utilize a recently proposed improved crystal graph convolutional neural network (iCGCNN) model, and the Voronoi tessellation approach incorporated in the Materials-Agnostic Platform for Informatics and Exploration (MAGPIE). ML models are trained on several training datasets prior calculated in the Open Quantum Materials Database (OQMD) and evaluated on a same separate test set of 380 mixed-anion compounds. Surprisingly, the ML model trained on 3,460 unrelaxed mixed-anion compounds only dataset outperforms the other models with MAE 0.116 eV/atom. We therefore make predictions on a separate ~4,000 hypothetical mixed-anion compounds, which are subsequently validated by DFT calculations. We find 51 new (meta)stable mixed anion compounds using only 241 DFT calculations, a success rate of 21.2%, more than 3 times what is achieved in a typical high-throughput survey.   

On Demand

Undesired dimerization of proteins has often been linked to a variety of diseases. This article studies high throughput screening of drug-like molecules capable of binding to Interferon Regulatory Factor (IRF) proteins in order to inhibit dimerization of these proteins. A population of small molecules with physicochemical characteristics needed for the inhibition of protein dimerization was generated by the development of a genetic algorithm by using RDkit. The molecule screening was performed by protein-ligand docking for the best binding affinity on proteins by using AutoDock Vina and Rosetta software. The screening outputs were validated by comparing them with experimental results.