Table of contents

Data science is playing a transformational role in all aspects of science, engineering, commerce, and society. Here we raise the questions, "what is data science, and why is it relevant to ECS?" We tackle these questions from several different perspectives: using definitions provided by thought leaders and early adopters; summarizing the four papers contributed by leading electrochemical and solid state researchers; and adopting a realpolitik understanding of the world we live in.

Recent advances in open source software capabilities, concurrent with advances in computational chemistry and operando diagnostics, provide new opportunities for modeling complex electrochemical phenomena. These capabilities include (i) new tools for modeling electrochemical processes; (ii) software for efficient and powerful numerical simulation, analysis, and visualization; (iii) and software ecosystem tools to enable efficient management and development of new computational tools. Herein, we highlight some of the open source software tools relevant to electrochemical simulations and demonstrate these capabilities within the context of modeling the growth of the solid electrolyte interphase in lithium ion batteries. Finally, we discuss the future outlook for open source software tools within the context of data science and electrochemistry research.

In simple half-cell electrochemical systems, straightforward models and analysis can be used for interpretation of experimental data. However, as the electrochemical systems we study become more complex and the mathematical models more sophisticated, the development of community-driven, open-source analysis tools and open-data become increasingly important vehicles for research reproducibility, integration of models with experimental data, and accelerated adoption of new methods and best practices. In this article, we focus on the development of an open-source tool for analyzing electrochemical impedance spectroscopy (EIS) data, and show how an automated workflow can enhance the integration of data from different instruments, while allowing flexible user-developed methods for data validation, creation and comparison of diverse models, and better model-data convergence through augmented statistical analysis and visualization capabilities. Given the breadth of application areas for EIS and the sophisticated insights it affords, we believe open-software and open-data are critical tool for accelerating electrochemical science and technology.

Despite recent rapid progress in lithium-ion batteries, cost and performance challenges remain. Systems engineering, including modelling, data analysis, and control, has a role to play in understanding and improving energy storage systems. For example, machine learning tools can be applied to estimate and predict battery degradation – the gradual decrease in capacity and increase in resistance that occurs as lithium-ion cells age. This is a two-fold challenge: first, estimation of the current state of health from field data in real applications, and second, prediction of the future evolution of state of health. Initial results are encouraging, but remain limited in scope and generality. Challenges such as efficient scaling up of algorithms and validation of predictions remain. For further breakthroughs, access to large data sets of battery usage over long periods of time are required, which requires the community to share data openly and in a transparent and portable manner.

Distributed computing, data science, and machine learning are producing transformative changes across diverse research areas. Our research focuses on increasing the lifetime performance of photovoltaic (PV) module, and is essential to increasing PV energy generation on the electrical grid. Traditional analysis of PV modules is insufficient to determine accurate lifetimes of modules with different architectures deployed in diverse climatic zones. To solve this complex problem, a data science approach is needed to handle the large scale data on materials, modules, commercial power plants, and the grid. This approach involves data ingestion with a non-relational data warehouse and data driven modeling based on the underlying physics and chemistry. It is critical to assemble data, develop and share codes and tools, and report research results to the whole PV value chain, as opposed to just the PV research community.