MLstructureMining: a machine learning tool for structure identification from X-ray pair distribution functions.

Synchrotron X-ray techniques are essential for studies of the intrinsic relationship between synthesis, structure, and properties of materials. Modern synchrotrons can produce up to 1 petabyte of data per day. Such amounts of data can speed up materials development, but also comes with a staggering growth in workload, as the data generated must be stored and analyzed. We present an approach for quickly identifying an atomic structure model from pair distribution function (PDF) data from (nano)crystalline materials. Our model, MLstructureMining, uses a tree-based machine learning (ML) classifier. MLstructureMining has been trained to classify chemical structures from a PDF and gives a top-3 accuracy of 99% on simulated PDFs not seen during training, with a total of 6062 possible classes. We also demonstrate that MLstructureMining can identify the chemical structure from experimental PDFs from nanoparticles of CoFe2O4 and CeO2, and we show how it can be used to treat an in situ PDF series collected during Bi2Fe4O9 formation. Additionally, we show how MLstructureMining can be used in combination with the well-known methods, principal component analysis (PCA) and non-negative matrix factorization (NMF) to analyze data from in situ experiments. MLstructureMining thus allows for real-time structure characterization by screening vast quantities of crystallographic information files in seconds.

POMFinder: identifying polyoxometallate cluster structures from pair distribution function data using explainable machine learning.

Characterization of a material structure with pair distribution function (PDF) analysis typically involves refining a structure model against an experimental data set, but finding or constructing a suitable atomic model for PDF modelling can be an extremely labour-intensive task, requiring carefully browsing through large numbers of possible models. Presented here is POMFinder, a machine learning (ML) classifier that rapidly screens a database of structures, here polyoxometallate (POM) clusters, to identify candidate structures for PDF data modelling. The approach is shown to identify suitable POMs from experimental data, including in situ data collected with fast acquisition times. This automated approach has significant potential for identifying suitable models for structure refinement to extract quantitative structural parameters in materials chemistry research. POMFinder is open source and user friendly, making it accessible to those without prior ML knowledge. It is also demonstrated that POMFinder offers a promising modelling framework for combined modelling of multiple scattering techniques.

ClusterFinder: a fast tool to find cluster structures from pair distribution function data.

A novel automated high-throughput screening approach, ClusterFinder, is reported for finding candidate structures for atomic pair distribution function (PDF) structural refinements. Finding starting models for PDF refinements is notoriously difficult when the PDF originates from nanoclusters or small nanoparticles. The reported ClusterFinder algorithm can screen 104 to 105 candidate structures from structural databases such as the Inorganic Crystal Structure Database (ICSD) in minutes, using the crystal structures as templates in which it looks for atomic clusters that result in a PDF similar to the target measured PDF. The algorithm returns a rank-ordered list of clusters for further assessment by the user. The algorithm has performed well for simulated and measured PDFs of metal-oxido clusters such as Keggin clusters. This is therefore a powerful approach to finding structural cluster candidates in a modelling campaign for PDFs of nanoparticles and nanoclusters.

Formation of intermetallic PdIn nanoparticles: influence of surfactants on nanoparticle atomic structure.

Bimetallic nanoparticles have been extensively studied as electrocatalysts due to their superior catalytic activity and selectivity compared to their monometallic counterparts. The properties of bimetallic materials depend on the ordering of the metals in the structure, and to tailor-make materials for specific applications, it is important to be able to control the atomic structure of the materials during synthesis. Here, we study the formation of bimetallic palladium indium nanoparticles to understand how the synthesis parameters and additives used influence the atomic structure of the obtained product. Specifically, we investigate a colloidal synthesis, where oleylamine was used as the main solvent while the effect of two surfactants, oleic acid (OA) and trioctylphosphine (TOP) was studied. We found that without TOP included in the synthesis, a Pd-rich intermetallic phase with the Pd3In structure initially formed, which transformed into large NPs of the CsCl-structured PdIn phase. When TOP was included, the syntheses yielded both In2O3 and Pd3In. In situ X-ray total scattering with Pair Distribution Function analysis was used to study the formation process of PdIn bimetallic NPs. Our results highlight how seemingly subtle changes to material synthesis methods can have a large influence on the product atomic structure.

Machine learning for analysis of experimental scattering and spectroscopy data in materials chemistry.

The rapid growth of materials chemistry data, driven by advancements in large-scale radiation facilities as well as laboratory instruments, has outpaced conventional data analysis and modelling methods, which can require enormous manual effort. To address this bottleneck, we investigate the application of supervised and unsupervised machine learning (ML) techniques for scattering and spectroscopy data analysis in materials chemistry research. Our perspective focuses on ML applications in powder diffraction (PD), pair distribution function (PDF), small-angle scattering (SAS), inelastic neutron scattering (INS), and X-ray absorption spectroscopy (XAS) data, but the lessons that we learn are generally applicable across materials chemistry. We review the ability of ML to identify physical and structural models and extract information efficiently and accurately from experimental data. Furthermore, we discuss the challenges associated with supervised ML and highlight how unsupervised ML can mitigate these limitations, thus enhancing experimental materials chemistry data analysis. Our perspective emphasises the transformative potential of ML in materials chemistry characterisation and identifies promising directions for future applications. The perspective aims to guide newcomers to ML-based experimental data analysis.

Atomic structural changes in the formation of transition metal tungstates: the role of polyoxometalate structures in material crystallization.

Material nucleation processes are poorly understood; nevertheless, an atomistic understanding of material formation would aid in the design of material synthesis methods. Here, we apply in situ X-ray total scattering experiments with pair distribution function (PDF) analysis to study the hydrothermal synthesis of wolframite-type MWO4 (M : Mn, Fe, Co, Ni). The data obtained allow the mapping of the material formation pathway in detail. We first show that upon mixing of the aqueous precursors, a crystalline precursor containing [W8O27]6- clusters forms for the MnWO4 synthesis, while amorphous pastes form for the FeWO4, CoWO4 and NiWO4 syntheses. The structure of the amorphous precursors was studied in detail with PDF analysis. Using database structure mining and an automated modelling strategy by applying machine learning, we show that the amorphous precursor structure can be described through polyoxometalate chemistry. A skewed sandwich cluster containing Keggin fragments describes the PDF of the precursor structure well, and the analysis shows that the precursor for FeWO4 is more ordered than that of CoWO4 and NiWO4. Upon heating, the crystalline MnWO4 precursor quickly converts directly to crystalline MnWO4, while the amorphous precursors transform into a disordered intermediate phase before the crystalline tungstates appear. Our data show that the more disordered the precursor is, the longer the reaction time required to form crystalline products, and disorder in the precursor phase appears to be a barrier for crystallization. More generally, we see that polyoxometalate chemistry is useful when describing the initial wet-chemical formation of mixed metal oxides.

Correction: Characterisation of intergrowth in metal oxide materials using structure-mining: the case of γ-MnO2.

Correction for 'Characterisation of intergrowth in metal oxide materials using structure-mining: the case of γ-MnO2' by Nicolas P. L. Magnard et al., Dalton Trans., 2022, 51, 17150-17161, https://doi.org/10.1039/D2DT02153F.

DeepStruc: towards structure solution from pair distribution function data using deep generative models.

Structure solution of nanostructured materials that have limited long-range order remains a bottleneck in materials development. We present a deep learning algorithm, DeepStruc, that can solve a simple monometallic nanoparticle structure directly from a Pair Distribution Function (PDF) obtained from total scattering data by using a conditional variational autoencoder. We first apply DeepStruc to PDFs from seven different structure types of monometallic nanoparticles, and show that structures can be solved from both simulated and experimental PDFs, including PDFs from nanoparticles that are not present in the training distribution. We also apply DeepStruc to a system of hcp, fcc and stacking faulted nanoparticles, where DeepStruc recognizes stacking faulted nanoparticles as an interpolation between hcp and fcc nanoparticles and is able to solve stacking faulted structures from PDFs. Our findings suggests that DeepStruc is a step towards a general approach for structure solution of nanomaterials.

Characterisation of intergrowth in metal oxide materials using structure-mining: the case of γ-MnO2.

Manganese dioxide compounds are widely used in electrochemical applications e.g. as electrode materials or photocatalysts. One of the most used polymorphs is γ-MnO2 which is a disordered intergrowth of pyrolusite ß-MnO2 and ramsdellite R-MnO2. The presence of intergrowth defects alters the material properties, however, they are difficult to characterise using standard X-ray diffraction due to anisotropic broadening of Bragg reflections. We here propose a characterisation method for intergrown structures by modelling of X-ray diffraction patterns and pair distribution functions (PDF) using γ-MnO2 as an example. Firstly, we present a fast peak-fitting analysis approach, where features in experimental diffraction patterns and PDFs are matched to simulated patterns from intergrowth structures, allowing quick characterisation of defect densities. Secondly, we present a structure-mining-based analysis using simulated γ-MnO2 superstructures which are compared to our experimental data to extract trends on defect densities with synthesis conditions. We applied the methodology to a series of γ-MnO2 samples synthesised by a hydrothermal route. Our results show that with synthesis time, the intergrowth structure reorders from a R-like to a ß-like structure, with the ß-MnO2 fraction ranging from ca. 27 to 82% in the samples investigated here. Further analysis of the structure-mining results using machine learning can enable extraction of more nanostructural information such as the distribution and size of intergrown domains in the structure. Using this analysis, we observe segregation of R- and ß-MnO2 domains in the manganese oxide nanoparticles. While R-MnO2 domains keep a constant size of ca. 1-2 nm, the ß-MnO2 domains grow with synthesis time.

We are never ever getting (back to) ideal symmetry: structure and luminescence in a ten-coordinated europium(III) sulfate crystal.

Our theoretical treatment of electronic structures in coordination complexes often rests on assumptions of symmetry. Experiments rarely provide fully symmetric systems to study. In solutions, fluctuations in solvation, variations in conformations, and even changes in constitution occur and complicate the picture. In crystals, lattice distortion, energy transfer, and phonon quenching play a role, but we are able to identify distinct symmetries. Yet the question remains: How is the real symmetry in a crystal compared to ideal symmetries?

Breaking with the Principles of Coreduction to Form Stoichiometric Intermetallic PdCu Nanoparticles.

Intermetallic nanoparticles (NPs) have shown enhanced catalytic properties as compared to their disordered alloy counterparts. To advance their use in green energy, it is crucial to understand what controls the formation of intermetallic NPs over alloy structures. By carefully selecting the additives used in NP synthesis, it is here shown that monodisperse, intermetallic PdCu NPs can be synthesized in a controllable manner. Introducing the additives iron(III) chloride and ascorbic acid, both morphological and structural control can be achieved. Combined, these additives provide a synergetic effect resulting in precursor reduction and defect-free growth; ultimately leading to monodisperse, single-crystalline, intermetallic PdCu NPs. Using in situ X-ray total scattering, a hitherto unknown transformation pathway is reported that diverges from the commonly reported coreduction disorder-order transformation. A Cu-rich structure initially forms, which upon the incorporation of Pd(0) and atomic ordering forms intermetallic PdCu NPs. These findings underpin that formation of stoichiometric intermetallic NPs is not limited by standard reduction potential matching and coreduction mechanisms, but is instead driven by changes in the local chemistry. Ultimately, using the local chemistry as a handle to tune the NP structure might open new opportunities to expand the library of intermetallic NPs by exploiting synthesis by design.

Simple Setup Miniaturization with Multiple Benefits for Green Chemistry in Nanoparticle Synthesis.

The development of nanomaterials often relies on wet-chemical synthesis performed in reflux setups using round-bottom flasks. Here, an alternative approach to synthesize nanomaterials is presented that uses glass tubes designed for NMR analysis as reactors. This approach uses less solvent and energy, generates less waste, provides safer conditions, is less prone to contamination, and is compatible with high-throughput screening. The benefits of this approach are illustrated by an in breadth study with the synthesis of gold, iridium, osmium, and copper sulfide nanoparticles.

Size-induced amorphous structure in tungsten oxide nanoparticles.

The properties of functional materials are intrinsically linked to their atomic structure. When going to the nanoscale, size-induced structural changes in atomic structure often occur, however these are rarely well-understood. Here, we systematically investigate the atomic structure of tungsten oxide nanoparticles as a function of the nanoparticle size and observe drastic changes when the particles are smaller than 5 nm, where the particles are amorphous. The tungsten oxide nanoparticles are synthesized by thermal decomposition of ammonium metatungstate hydrate in oleylamine and by varying the ammonium metatungstate hydrate concentration, the nanoparticle size, shape and structure can be controlled. At low concentrations, nanoparticles with a diameter of 2-4 nm form and adopt an amorphous structure that locally resembles the structure of polyoxometalate clusters. When the concentration is increased the nanoparticles become elongated and form nanocrystalline rods up to 50 nm in length. The study thus reveals a size-dependent amorphous structure when going to the nanoscale and provides further knowledge on how metal oxide crystal structures change at extreme length scales.

Structural Changes during the Growth of Atomically Precise Metal Oxido Nanoclusters from Combined Pair Distribution Function and Small-Angle X-ray Scattering Analysis.

The combination of in situ pair distribution function (PDF) analysis and small-angle X-ray scattering (SAXS) enables analysis of the formation mechanism of metal oxido nanoclusters and cluster-solvent interactions as they take place. Herein, we demonstrate the method for the formation of clusters with a [Bi38 O45 ] core. Upon dissolution of crystalline [Bi6 O5 (OH)3 (NO3 )5 ]⋅3 H2 O in DMSO, an intermediate rapidly forms, which slowly grows to stable [Bi38 O45 ] clusters. To identify the intermediate, we developed an automated modeling method, where smaller [Bix Oy ] structures based on the [Bi38 O45 ] framework are tested against the data. [Bi22 O26 ] was identified as the main intermediate species, illustrating how combined PDF and SAXS analysis is a powerful tool to gain insight into nucleation on an atomic scale. PDF also provides information on the interaction between nanoclusters and solvent, which is shown to depend on the nature of the ligands on the cluster surface.

Formation and growth mechanism for niobium oxide nanoparticles: atomistic insight from in situ X-ray total scattering.

Understanding the mechanisms for nanoparticle nucleation and growth is crucial for the development of tailormade nanomaterials. Here, we use X-ray total scattering and Pair Distribution Function analysis to follow the formation and growth of niobium oxide nanoparticles. We study the solvothermal synthesis from niobium chloride in benzyl alcohol, and through investigations of the influence of reaction temperature, a formation pathway can be suggested. Upon dissolution of niobium chloride in benzyl alcohol, octahedral [NbCl6-xOx] complexes form through exchange of chloride ligands. Heating of the solution results in polymerization, where larger clusters built from multiple edge-sharing [NbCl6-xOx] octahedra assemble. This leads to the formation of a nucleation cluster with the ReO3 type structure, which grows to form nanoparticles of the Wadsley-Roth type H-Nb2O5 structure, which in the bulk phase usually only forms at high temperature. Upon further growth, structural defects appear, and the presence of shear-planes in the structure appears highly dependent on nanoparticle size.