07-10, 13:55–14:25 (US/Pacific), Room 316
Johnson Matthey (JM) leads in sustainable technologies, employing advanced science to address global challenges in energy, chemicals, and automotive sectors. Our cutting-edge research and development (R&D) facilities include state-of-the-art characterization tools, handling diverse datasets like images, timeseries, 3D tomograms, spectra, and digital twins. With the rising demand for data-driven insights, Python has emerged as a vital tool in enhancing decision-making processes. We showcase our utilization of the open-source community to construct our data science research platform, marking a significant step forward in our innovation journey.
Johnson Matthey stands as a global leader in sustainable technologies, employing advanced science to address critical challenges in energy, chemicals, and automotive industries. Beyond aesthetic appeal, microscopy serves a pivotal role in understanding atomic-level processes, crucial for comprehending oxygen binding mechanisms in nanoparticles. Our commitment, spanning six generations of PhD students and acquisition of an aberration-corrected electron microscope, underscores JM's dedication to catalysis science. These endeavours not only enrich our understanding but also democratize access to catalysis insights within JM.
This presentation delves into the intersection of microscopic observations and catalytic activity, elucidating instrumentation techniques that enable quantitative image capture and the generation of digital twins of our materials. Rapid characterization of materials requires machine learning and automation. Leveraging Python and its open-source community, we utilize libraries such as Dask, Dash, iPyvolume, Scikit-image, PyTorch, and more in unison to characterize our materials. Bridging length scales necessitates a correlative approach, demonstrated through streamlined data processing pipelines. The approach to characterization highlights the significance of integrated tools developed from the ground up for image segmentation, visualization, particle metrology, and data preparation for simulations. Collaborative data-driven characterizations benefit JM holistically, showcasing applications in clean air and fuel cells research projects.
Leveraging recent front-end frameworks in Python, we have constructed an apps platform, the first of its kind in the chemicals industry. This platform facilitates seamless collaboration and access to analysis tools across JM, integrating high-performance compute Python libraries to scale calculations efficiently and handle vast datasets with ease. By harnessing the power of these tools, we streamline data processing pipelines and enhance our capability to tackle complex research challenges across multiple length scales.
Looking towards the future, we demonstrate the start of techniques for live data processing from our microscopes. We envision a future where live data acquisition from microscopes transcends mere observation, empowering a feedback loop to drive microscope operations autonomously. This iterative process, enriched by real-time structural inversion techniques, enables the creation of accurate structural models on the fly. By integrating these models into simulations, we can dynamically adjust experimental parameters, optimizing catalyst performance and advancing material design. This seamless integration of live data acquisition, feedback-driven microscopy, and structural inversion represents a paradigm shift, accelerating our journey towards predictive materials design and revolutionizing the landscape of catalysis science.
JM remains dedicated to pushing the boundaries of sustainable technology, using cutting-edge science to tackle real-world problems in energy, chemicals, and automotive sectors. By combining advanced tools and collaborative research, we unlock new insights that benefit everyone within the organization. With our latest ventures into developing an apps platform and live data processing from microscopes, we take a significant step toward predicting material behaviour, revolutionizing how we approach catalysis science with Python and open-source communities.
Aakash Varambhia works as a Data Science Lead at Johnson Matthey, focusing on delivering innovative imaging data science solutions for research and development. He completed his DPhil at the University of Oxford, specializing in using advanced tools for quantitative electron microscopy. His expertise covers analyzing various data types like images, timeseries, and spectra, all aimed at improving materials design processes.