In Situ Studies of the Formation of Tungsten and Niobium Oxide Nanoparticles: Towards Automated Analysis of Reaction Pathways from PDF Analysis using the Pearson Correlation Coefficient

Abstract

Invited for this month's cover are the groups of Kirsten M. Ø. Jensen at the University of Copenhagen (Denmark) and Simon J. L. Billinge at Columbia University (US). The cover picture shows a sketch of the formation process of metal oxide nanoparticles, where nanocrystalline oxides form from fragments of e.g., polyoxometalates. In situ X‐ray total scattering studies with Pair Distribution Function analysis can give new insights into the formation process, as it provides structural information on all stages of the reaction – from precursor ions in solution, over amorphous or nanostructured intermediates to the final crystalline material. Here, it is show how the analysis of such data can be automated using structure mining and simple computational tools. Read the full text of their Research Article at 10.1002/cmtd.202200034.

In one word, how would you describe your research?

[IMAGE OMITTED. SEE PDF.]

We will use two words – Nanostructure analysis. The research done in our research group at University of Copenhagen focuses on understanding synthesis/structure and structure/property relations in nanostructured, inorganic materials. For this work, we use a range of X-ray and neutron methods, especially X-ray total scattering with Pair Distribution Function (PDF) analysis. In contrast to other crystallographic methods, PDF analysis allows to extract structural information from materials with no long-range order, which makes it possible to develop detailed models for the atomic arrangements in nanomaterials. We use X-ray total scattering methods for e.g., in situ studies of material formation, which allows us to map reaction pathways and link reaction mechanism to polymorph formation. We collaborate with the Billinge group at Columbia University, who are leading in development of the PDF method.

What prompted you to investigate this topic/problem?

Over the past years, we have done many time- and position resolved in situ scattering studies at synchrotron facilities around the world. Such experiments often yield large sets of data, and the analysis of the data is quite time consuming and challenging. One needs to carefully analyse the changes that appear with e.g., reaction time or sample position, and it is often necessary to test many structural models against the experimental data to find one that is suitable for further analysis. We thus had an interest in developing simple tools that could help in this work. Here, we first show that simple computational tools based on the Pearson correlation coefficient can be used to map the structural changes that take place during a reaction, so that the most interesting data points can be easily identified. We then show that the Pearson correlation coefficient is also useful for ‘structure mining’, where a large number of models are automatically compared to experimental data, so that a good structure model can be found.

Is your current research mainly curiosity driven (fundamental) or rather applied?

Many of our projects concern very fundamental questions in materials chemistry: How does a material form, and how can we control its atomic and nanoscale structure? How does nanosize affect the atomic structure of a material, and how does it affect its properties? What are the structural changes that take place during e.g., use or degradation of functional materials? However, one of the wonderful things about working in materials chemistry is that the distance between fundamental and applied research is often quite short, as these questions are also relevant for development of applied, functional materials for a greener future. [IMAGE OMITTED. SEE PDF.]