Lithium ion batteries emit very faint sounds as they charge, discharge, and degrade, but until now no one could interpret those sounds to detect when a battery might be about to lose power, fail, or burst into flames. Mit engineers have now demonstrated a practical way to decode those signals, even when they are buried in noisy data. Their findings point to the possibility of relatively simple, totally passive, and nondestructive monitoring devices that could track the health of batteries in electric vehicles and grid scale storage facilities in real time.
The research team, led by professor of chemical engineering and mathematics Martin Z. Bazant, focused on identifying how different degradation mechanisms imprint themselves in a battery’s acoustic emissions. Bazant explains that “through some careful scientific work, our team has managed to decode the acoustic emissions,” and they were able to classify these emissions as originating either from gas bubbles generated by side reactions or from fractures caused by expansion and contraction of the active material. These two signatures correspond to primary pathways of battery degradation and failure, giving engineers a way to distinguish between them using sound alone.
To build that mapping from sound to failure mode, the engineers coupled electrochemical testing of working batteries with recordings of their acoustic emissions, using signal processing to link specific sound characteristics with voltage and current behavior over time. After testing, they disassembled the batteries and examined them with an electron microscope to directly observe fracturing and validate the acoustic signals. In collaboration with Oak Ridge National Laboratory researchers, they further showed that acoustic emissions can provide early warning of gas generation before thermal runaway, a process that can lead to fires. Bazant likens this early warning to “seeing the first tiny bubbles in a pot of heated water, long before it boils,” suggesting that continuous acoustic monitoring could catch dangerous failure modes well in advance.
