The power of ultrasound & machine learning
Lithium-ion batteries have transformed consumer electronics and communication technology by putting computers in everyone’s hands. Now, they are transforming vehicles and transportation.
Due to accelerating commitments from automakers to electrify their portfolios, the battery industry is scaling at never seen before rates. To meet sustainability targets, battery cell production needs to grow by ten times its current volume this decade while reducing the cost by more than 40 percent, without sacrificing quality and consistency.
A battery pack is only as strong as its weakest cell
Building a battery factory is hard. Even for experienced manufacturers, it can take a full year from the start of production to achieve 50% yield and another year or two to reach high enough yield and reliability levels for profitability.
There has been a massive gap in battery inspection and intelligence during battery production, especially in cell finishing, when battery cells have been fully assembled and are impervious to visual inspection. The market has relied to date on electrical test data, which inherently only provides cell-averaged information, overlooks subtle variations in large EV cells and has limited ability to detect process issues and predict quality during cell production.
Desperate for increased performance, automotive OEMs frequently build extra cells into battery packs (10% percent more than necessary) to reduce the risk of low performance later in their lifetime. Electrical methods fail to uncover damaged cells because even if the cells all respond the same during testing at the beginning of life, significant deviations in performance can develop later in life. This overbuilding results in a heavier, more costly battery pack and the construction of more manufacturing sites to produce these cells.
Electrical methods are less effective in EV-sized cells
Undetected process issues cause poor in-field reliability
The cells in a battery pack are interconnected and only perform at the weakest cell’s overall output. It is costly when underperforming or defective batteries are detected after leaving the factory. Identifying production issues as early as possible decreases the cost of those issues.
EchoStat® at work
EchoStat identifies issues earlier in the production process when the error occurs. Assessing poor cells earlier reduces scrap and costs while increasing the overall quality of cell production. When anomalies happen, EchoStat identifies outliers throughout battery production and when it matters most.
Liminal’s solutions tighten the feedback loop between error and detection in battery manufacturing.
Wetting & Soaking
The electrolyte is the lifeblood of a battery, as it transports lithium ions throughout the cell and is what allows the electrodes to store energy. Proper and consistent saturation of electrolytes is a crucial part of cell performance. It is a critical process to get right in production.
The challenge is that when battery cells are fully assembled and the electrolyte is injected into the cell, there is not currently a non-destructive method to detect whether the electrolyte has been adequately distributed. It is a literal black box at scale. Until now, there was no method for capturing the subtle variations of electrolyte distribution. Poor electrolyte distribution can be extremely significant to production quality and long-term performance.
EchoStat uses proprietary ultrasonic and analytics technology to reveal precisely when the cells reach complete saturation levels. During process development, Liminal customers have tested and reduced the soaking process by up to 50 percent in most cases – for example, reducing soak time from 24 to 12 hours to reach full saturation with no change in cell quality. The results are faster production time, higher equipment utilization rates, and lower cell cost.
Ultrasound detects full
saturation ~ 12 hours after
fill, not 24 hours
Lower cost of scrap, and ability to recycle or rework
More efficient processes, faster cell build times, and higher throughput
Soak less, reveal more
Case Study #1
In the battery manufacturing process, the manufacturer used a particular electrolyte until they ran out. The manufacturer then switched to a new batch from the same vendor that was meant to have identical composition to the first batch of electrolyte.
Five hours after the cells were filled, the EchoStat discovered saturation differences between the initial batch and the new batch. Running in parallel with EchoStat, the standard electrical testing did not detect discrepancies between batches. The cells were subsequently fast cycled and the first batch of 24 cells had a longer life than the second batch. This detection by EchoStat enabled the manufacturer to investigate and address the root cause of the discrepancy.