Automated Classification of Big X-ray Diffraction Data Using Deep Learning Models

Abstract Current in-situ X-ray diffraction (XRD) techniques generate data over human analytical capabilities – leading to the loss of novel insights. Automated techniques require human intervention, and lack the performance and adaptability needed for material exploration. With the critical need for high-throughput automated XRD pattern analysis of novel materials, we developed a generalized deep learning model to classify a diverse set of materials’ crystal systems and space groups. In our approach, we generated training data with a holistic representation of patterns that emerge from varying experimental conditions and crystal properties. We then used an expedited learning technique to refine our model’s expertise to experimental conditions. Additionally, we used evaluation data to interpret our model’s decision-making and optimized model architecture to elicit classification based on Bragg’s Law. We evaluated our models on experimental data, novel materials, and altered cubic crystals, where we observed state-of-the-art performance and even greater advances in space group classification.