Multi-dimensional Data Visualization for Analyzing Materials

High-throughput chemical synthesis is extensively used for analyzing materials since it allows a systematic probing of a large span of three space parameters: initial conditions, final products, and the characteristics of the obtained products. High-dimensional data visualization is required to fully understand the relations between these three spaces. However, most methods are limited to up to three dimensions (3D) at the time per graph. Here, we show how correlation analysis and parallel coordinate plots reveal the relations in a multidimensional space. This representation technique is general and may serve many synthetic and fabrication related processes. We demonstrate the power of our approach on a specific chemical colloidal synthesis of CsPbBr3 nanocrystals, which results in highly emissive semiconductor nanocrystals, relevant for photonic applications. The colloidal synthesis of these nanocrystals is a multi-parameter process, with complex inter-relationships between the parameters, such as precursors and organic ligands concentrations and reaction temperature. The resulting nanocrystals present distinct morphological differences, resulting in a detectable shift of their emission spectrum. We use a dataset of 1351 samples to investigate the relations between and within these three spaces. In our case study, we have identified trends and anomalies in the data that provide directions for further research and illustrated thereby the potential of correlation analysis and parallel coordinates to explore patterns and relationships.