Search

Machine Learning (ML) has offered innovative perspectives for accelerating the discovery of new functional materials, leveraging the increasing availability of material databases. Despite the promising advances, data-driven methods face constraints imposed by the quantity and quality of available data. Moreover, ML is often employed in tandem with simulated datasets originating from density functional theory (DFT), and assessed through in-sample evaluation schemes. This scenario raises questions about the practical utility of ML in uncovering new and significant material classes for industrial applications. Here, we propose a data-driven framework aimed at accelerating the discovery of new transparent conducting materials (TCMs), an important category of semiconductors with a wide range of applications. To mitigate the shortage of available data, we create and validate unique experimental databases, comprising several examples of existing TCMs. We assess state-of-the-art (SOTA) ML models for property prediction from the stoichiometry alone. We propose a bespoke evaluation scheme to provide empirical evidence on the ability of ML to uncover new, previously unseen materials of interest. We test our approach on a list of 55 compositions containing typical elements of known TCMs. Although our study indicates that ML tends to identify new TCMs compositionally similar to those in the training data, we empirically demonstrate that it can highlight material candidates that may have been previously overlooked, offering a systematic approach to identify materials that are likely to display TCMs characteristics.

The scarcity of property labels remains a key challenge in materials informatics, whereas materials data without property labels are abundant in comparison. By pretraining supervised property prediction models on self-supervised tasks that depend only on the "intrinsic information" available in any Crystallographic Information File (CIF), there is potential to leverage the large amount of crystal data without property labels to improve property prediction results on small datasets. We apply Deep InfoMax as a self-supervised machine learning framework for materials informatics that explicitly maximises the mutual information between a point set (or graph) representation of a crystal and a vector representation suitable for downstream learning. This allows the pretraining of supervised models on large materials datasets without the need for property labels and without requiring the model to reconstruct the crystal from a representation vector. We investigate the benefits of Deep InfoMax pretraining implemented on the Site-Net architecture to improve the performance of downstream property prediction models with small amounts (<10^3) of data, a situation relevant to experimentally measured materials property databases. Using a property label masking methodology, where we perform self-supervised learning on larger supervised datasets and then train supervised models on a small subset of the labels, we isolate Deep InfoMax pretraining from the effects of distributional shift. We demonstrate performance improvements in the contexts of representation learning and transfer learning on the tasks of band gap and formation energy prediction. Having established the effectiveness of Deep InfoMax pretraining in a controlled environment, our findings provide a foundation for extending the approach to address practical challenges in materials informatics.

In the last few decades, basic ideas of topology have completely transformed the prediction of quantum transport phenomena. Following this trend, we go deeper into the incorporation of modern mathematics into quantum material science focusing on geometry. Here we investigate the relation between the geometrical type of the Fermi surface and Anomalous and Spin Hall Effects. An index, $\mathbb{H}_F$, quantifying the hyperbolic geometry of the Fermi surface, shows a universal correlation (R$^2$ = 0.97) with the experimentally measured intrinsic anomalous Hall conductivity, of 16 different compounds spanning a wide variety of crystal, chemical, and electronic structure families, including those where topological methods give R$^2$ = 0.52. This raises a question about the predictive limits of topological physics and its transformation into a wider study of bandstructures' and Fermi surfaces' geometries, opening horizon for prediction of phenomena beyond topological understanding., Comment: Submission to SciPost. arXiv admin note: substantial text overlap with arXiv:2007.05525

Clustering is a common task in machine learning, but clusters of unlabelled data can be hard to quantify. The application of clustering algorithms in chemistry is often dependant on material representation. Ascertaining the effects of different representations, clustering algorithms, or data transformations on the resulting clusters is difficult due to the dimensionality of these data. We present a thorough analysis of measures for isotropy of a cluster, including a novel implantation based on an existing derivation. Using fractional anisotropy, a common method used in medical imaging for comparison, we then expand these measures to examine the average isotropy of a set of clusters. A use case for such measures is demonstrated by quantifying the effects of kernel approximation functions on different representations of the Inorganic Crystal Structure Database. Broader applicability of these methods is demonstrated in analysing learnt embedding of the MNIST dataset. Random clusters are explored to examine the differences between isotropy measures presented, and to see how each method scales with the dimensionality. Python implementations of these measures are provided for use by the community., Comment: 31 pages, 6 figures

Computational exploration of the compositional spaces of materials can provide guidance for synthetic research and thus accelerate the discovery of novel materials. Most approaches employ high-throughput sampling and focus on reducing the time for energy evaluation for individual compositions, often at the cost of accuracy. Here, we present an alternative approach focusing on effective sampling of the compositional space. The learning algorithm PhaseBO optimizes the stoichiometry of the potential target material while improving the probability of and accelerating its discovery without compromising the accuracy of energy evaluation.

Site-Net is a transformer architecture that models the periodic crystal structures of inorganic materials as a labelled point set of atoms and relies entirely on global self-attention and geometric information to guide learning. Site-Net processes standard crystallographic information files to generate a large real-space supercell, and the importance of interactions between all atomic sites is flexibly learned by the model for the prediction task presented. The attention mechanism is probed to reveal Site-Net can learn long-range interactions in crystal structures, and that specific attention heads become specialized to deal with primarily short- or long-range interactions. We perform a preliminary hyperparameter search and train Site-Net using a single graphics processing unit (GPU), and show Site-Net achieves state-of-the-art performance on a standard band gap regression task., Comment: 23 pages, 13 figures

Fundamental differences between materials originate from the unique nature of their constituent chemical elements. Before specific differences emerge according to the precise ratios of elements in a given crystal structure, a material can be represented by the set of its constituent chemical elements. By working at the level of the periodic table, assessment of materials at the level of their phase fields reduces the combinatorial complexity to accelerate screening, and circumvents the challenges associated with composition-level approaches such as poor extrapolation within phase fields, and the impossibility of exhaustive sampling. This early stage discrimination combined with evaluation of novelty of phase fields aligns with the outstanding experimental challenge of identifying new areas of chemistry to investigate, by prioritising which elements to combine in a reaction. Here, we demonstrate that phase fields can be assessed with respect to the maximum expected value of a target functional property and ranked according to chemical novelty. We develop and present PhaseSelect, an end-to-end machine learning model that combines the representation, classification, regression and ranking of phase fields. First, PhaseSelect constructs elemental characteristics from the co-occurrence of chemical elements in computationally and experimentally reported materials, then it employs attention mechanisms to learn representation for phase fields and assess their functional performance. At the level of the periodic table, PhaseSelect quantifies the probability of observing a functional property, estimates its value within a phase field and also ranks a phase field novelty, which we demonstrate with significant accuracy for three avenues of materials applications for high-temperature superconductivity, high-temperature magnetism, and targeted bandgap energy.

Neutron diffraction and X-ray pair distribution function (XPDF) experiments were performed in order to investigate the magnetic and local crystal structures of Ba2FeSbSe5 and to compare them to the average (i.e. long-range) structural model, previously obtained by single crystal X-ray diffraction. Changes in the local crystal structure (i.e. in the second coordination sphere) are observed upon cooling from 295 K to 95 K resulting in deviations from the average (i.e. long-range) crystal structure. This work demonstrates, that these observations cannot be explained by local or long-range magnetoelastic effects involving Fe-Fe correlations. Instead, we found, that the observed differences between local and average crystal structure can be explained by Sb-5s lone pair dynamics. We also find, that below the N\'eel temperature (TN = 58 K), the two distinct magnetic Fe3+ sites order collinearly, such that a combination of antiparallel and parallel spin arrangements along the b-axis results. The nearest-neighbor arrangement (J1 = 6 {\AA}) is fully antiferromagnetic, while next-nearest-neighbor interactions are ferromagnetic in nature.

Motivated by the famous and pioneering mathematical works by Perelman, Hamilton, and Thurston, we introduce the concept of using modern geometrical mathematical classifications of multi-dimensional manifolds to characterize electronic structures and predict non-trivial electron transport phenomena. Here we develop the Fermi Surface Geometry Effect (FSGE), using the concepts of tangent bundles and Gaussian curvature as an invariant. We develop an index, $\mathbb{H}_F$, for describing the the "hyperbolicity" of the Fermi Surface (FS) and show a universal correlation (R$^2$ = 0.97) with the experimentally measured intrinsic anomalous Hall effect of 16 different compounds spanning a wide variety of crystal, chemical, and electronic structure families, including where current methods have struggled. This work lays the foundation for developing a complete theory of geometrical understanding of electronic (and by extension magnonic and phononic) structure manifolds, beginning with Fermi surfaces. In analogy to the broad impact of topological physics, the concepts begun here will have far reaching consequences and lead to a paradigm shift in the understanding of electron transport, moving it to include geometrical properties of the E vs k manifold as well as topological properties., Comment: 9 pages, 4 figures

Low-dimensional magnetic materials with spin-$\frac{1}{2}$ moments can host a range of exotic magnetic phenomena due to the intrinsic importance of quantum fluctuations to their behavior. Here, we report the structure, magnetic structure and magnetic properties of copper(II) thiocyanate, Cu(NCS)$_2$, a one-dimensional coordination polymer which displays low-dimensional quantum magnetism. Magnetic susceptibility, electron paramagnetic resonance (EPR) spectroscopy, $^{13}$C magic-angle spinning nuclear magnetic resonance (MASNMR) spectroscopy, and density functional theory (DFT) investigations indicate that Cu(NCS)$_2$ behaves as a two-dimensional array of weakly coupled antiferromagnetic spin chains ($J_2 = 133(1)$ K, $\alpha = J_1/J_2 = 0.08$). Powder neutron-diffraction measurements confirm that Cu(NCS)$_2$ orders as a commensurate antiferromagnet below $T_\mathrm{N} = 12$ K, with a strongly reduced ordered moment (0.3 $\mu_\mathrm{B}$) due to quantum fluctuations., Comment: 11 pages, 7 figures

The experimental search for new thermoelectric materials remains largely confined to a limited set of successful chemical and structural families, such as chalcogenides, skutterudites, and Zintl phases. In principle, computational tools such as density functional theory (DFT) offer the possibility of rationally guiding experimental synthesis efforts toward very different chemistries. However, in practice, predicting thermoelectric properties from first principles remains a challenging endeavor, and experimental researchers generally do not directly use computation to drive their own synthesis efforts. To bridge this practical gap between experimental needs and computational tools, we report an open machine learning-based recommendation engine (http://thermoelectrics.citrination.com) for materials researchers that suggests promising new thermoelectric compositions, and evaluates the feasibility of user-designed compounds. We show that this engine can identify interesting chemistries very different from known thermoelectrics. Specifically, we describe the experimental characterization of one example set of compounds derived from our engine, RE12Co5Bi (RE = Gd, Er), which exhibits surprising thermoelectric performance given its unprecedentedly high loading with metallic d and f block elements, and warrants further investigation as a new thermoelectric material platform., Comment: 8 pages, 4 figures

A structural phase transition from cubic $Fd\bar{3}m$ to tetragonal $I$4$_1$/$amd$ symmetry with $c/a >$ 1 is observed at $T_{\rm{S}}$ = 16 K in spinel GeCo$_2$O$_4$ below the N\'eel temperature $T_N$ = 21 K. Structural and magnetic ordering appear to be decoupled with the structural distortion occurring at 16 K while magnetic order occurs at 21 K as determined by magnetic susceptibility and heat capacity measurements. An elongation of CoO$_6$ octahedra is observed in the tetragonal phase of GeCo$_2$O$_4$. We present the complete crystallographic description of GeCo$_2$O$_4$ in the tetragonal $I$4$_1$/$amd$ space group and discuss the possible origin of this distortion in the context of known structural transitions in magnetic spinels. GeCo$_2$O$_4$ exhibits magnetodielectric coupling below $T_{\rm{N}}$. The related spinels GeFe$_2$O$_4$ and GeNi$_2$O$_4$ have also been examined for comparison. Structural transitions were not detected in either compound down to $T \approx$ 8 K. Magnetometry experiments reveal in GeFe$_2$O$_4$ a second antiferromagnetic transition, with $T_{\rm{N1}}$ = 7.9 K and $T_{\rm{N2}}$ = 6.2 K, that was previously unknown, and that bear a similarity to the magnetism of GeNi$_2$O$_4$.

Magnetic ordering in the geometrically frustrated magnetic oxide spinels MgCr_2O_4 and ZnCr_2O_4 is accompanied by a structural change that helps relieve the frustration. Analysis of high-resolution synchrotron X-ray scattering reveals that the low-temperature structures are well described by a two-phase model of tetragonal I4_1/amd and orthorhombic Fddd symmetries. The Cr_4 tetrahedra of the pyrochlore lattice are distorted at these low-temperatures, with the Fddd phase displaying larger distortions than the I4_1/amd phase. The spin-Jahn-Teller distortion is approximately one order of magnitude smaller than is observed in first-order Jahn-Teller spinels such as NiCr_2O_4 and CuCr_2O_4. In analogy with NiCr_2O_4 and CuCr_2O_4, we further suggest that the precise nature of magnetic ordering can itself provide a second driving force for structural change.

Polycrystalline Nd2Ru2O7 samples have been prepared and examined using a combination of structural, magnetic, and electrical and thermal transport studies. Analysis of synchrotron X-ray and neutron diffraction patterns suggests some site disorder on the A-site in the pyrochlore sublattice: Ru substitutes on the Nd-site up to 7.0(3)%, regardless of the different preparative conditions explored. Intrinsic magnetic and electrical transport properties have been measured. Ru 4d spins order antiferromagnetically at 143 K as seen both in susceptibility and specific heat, and there is a corresponding change in the electrical resistivity behaviour. A second antiferromagnetic ordering transition seen below 10 K is attributed to ordering of Nd 4f spins. Nd2Ru2O7 is an electrical insulator, and this behaviour is believed to be independent of the Ru-antisite disorder on the Nd site. The electrical properties of Nd2Ru2O7 are presented in the light of data published on all A2Ru2O7 pyrochlores, and we emphasize the special structural role that Bi3+ ions on the A-site play in driving metallic behaviour. High-temperature thermoelectric properties have also been measured. When considered in the context of known thermoelectric materials with useful figures-of-merit, it is clear that Nd2Ru2O7 has excessively high electrical resistivity which prevents it from being an effective thermoelectric. A method for screening candidate thermoelectrics is suggested., Comment: 19 pages, 10 figures