This presentation discusses leveraging machine learning and robotic experimentation to accelerate innovation in battery materials. It explores how these advanced techniques streamline material discovery, optimize properties, and expedite the development of next-generation battery technologies.

In this talk, we will discuss how new computational approaches enabled by high-performance computing and machine learning algorithms are accelerating the traditional materials design and commercialization process for battery materials. As a case study, we present our recent positive results on a new, record-breaking, solid Li-ion conductor material Li8B10S19, which embodies the new data-driven R&D paradigm of machine learning–based discovery and human-based synthesis and scale-up.

All engineering companies want to build high-quality products efficiently and quickly. When developing new electric vehicles, OEMs consistently face major challenges with battery testing; the need to maximize range and charging efficiency can present blockers to rapid development, and remove OEMs’ competitive advantage. OEMs desperately need new ways to accelerate product development. Using AI-guided battery testing, they can change the game—reducing testing, enhancing learning, and maximizing product quality to supercharge innovation.

High-energy active materials are actively incorporated in commercial cells to meet the requirement for increased energy density in battery systems. Silicon-based active material's distinctive characteristics, marked by voltage hysteresis and volumetric expansion, result in unique challenges in automotive applications. This presentation will share emerging challenges associated with the implementation of high silicon content cells in automotive battery systems, and discuss solutions to these challenges that utilize multiphysics modeling and insights to improve BMS predictions.

The presentation will cover the difficulties and prospects of the e-mobility industry and how battery health monitoring can help maintain the quality, security, and durability of batteries in electric vehicles and other uses. It will be shown how data analytics and artificial intelligence can enhance battery design, testing, and management, and the goal of a comprehensive approach that considers the whole value chain and life cycle of batteries, from raw materials to recycling.

Nissan, the company that launched the world’s first mass-produced electric vehicle, announced its goal to achieve carbon neutrality by 2050 throughout the vehicle’s lifecycle in 2021. To achieve this goal, there four challenges 1) Data-driven chemistry design, 2) Material recycle, 3) Cell design optimization, 4) Battery diagnosis/prognosis. This presentation will provide an overview of our approach using materials informatics technology in data-driven chemistry design.

Optimal deployment and accelerated development of batteries requires a deep understanding of battery performance and health alongside methods to predict the evolution of these quantities with respect to historical and anticipated stressors. In this presentation, we discuss the recontextualization of diverse health metrics as an advanced state of health (A-SOH) and introduce deep learning algorithms that show promise in predicting the future A-SOH for both real and simulated datasets.

Rapid electrochemical diagnostics, like DC pulse sequences or electrochemical impedance spectroscopy, are known to be useful for capacity prediction, but it is still unclear if they are useful in practice. To that end, NREL has collected a data set with ~50,000 DC pulses from four types of commercial lithium-ion batteries to enable training machine-learning models predicting state-of-charge/health/safety. We demonstrate that rapid DC pulses can be used to predict capacity with 2%-9% average error, which can separate high- from low-capacity cells but is not accurate enough to estimate remaining useful life, but that we cannot predict safety relevant targets.

Cell degradation is commonly characterized and quantified by various electrochemical signals. However, there is more to the story than meets the eye. We investigate the impact of aging parameters on cell internal changes characterized through high-resolution, high-throughput CT scanning with the goal of improving cell performance and durability.

Battery characterization and aging tests typically span several months to years, posing significant challenges for manufacturers and OEMs seeking to accelerate testing and extract comprehensive insights, particularly on battery aging. This work addresses these challenges by integrating physical modeling with machine learning to analyze battery performance at the parameter level. Leveraging robotics and high-throughput testing platforms for commercial cells, we develop and validate a framework that digitalizes and automates the testing process, enabling faster, more efficient, and data-driven battery evaluation.

Pseudo-Electrochemical Impedance Spectroscopy (pseudo-EIS) is a technique used to estimate the state of health of lithium-ion batteries (LIBs). Unlike EIS, pseudo-EIS can be implemented in a battery management system for real-time battery health diagnostics. This study uses pseudo-EIS applied to large-format pouch-type LIBs subject to both calendar and cycle life aging. The results correlated capacity and resistance changes with degradation mechanisms and the evolution of the parameters with aging.

This presentation explores the integration of active battery management systems in Li-metal batteries, enhancing safety and performance. It discusses innovations in management strategies, including monitoring, regulation, and fault detection, crucial for advancing the reliability and efficiency of next-generation energy storage solutions.