To subscribe to the Battery Smarts monthly newsletter, please click here. We cover a variety of battery storage and manufacturing topics with no advertising, cancel anytime.
When we speak to customers about offering full service analytics around batteries, they often think about one of two things: collecting battery cycler data in an R&D or manufacturing context and analyzing it, or collecting grid-connected battery data and analyzing that. While we do both regularly, we also often provide “data lifecycle intelligence” for manufacturers of battery components upstream of these use cases. This may include manufacturers of cathode and anode materials, electrolytes, foils and binders. Such scenarios, while they involve different materials and processes, are surprisingly well-suited to serialized data threading and introduce many of the same benefits that manufacturers and operators enjoy.
As most readers will already know, the number of research efforts pursuing novel approaches to lithium-ion (Li-ion) battery design has grown substantially in recent years, driven largely by ballooning demand for new technology uses in transportation and energy storage. Research has centered around solving challenges around energy density, as well as improving safety, cycle durability and lowered cost. Such efforts will only become more important as battery manufacturing in the U.S. will grow to more than $400 billion by 2030.1
A tremendous amount of attention has been paid to anode design, whether it’s solid Li or improvements to carbon-only anodes (using SiC instead of only C is an approach with well-known technical challenges) or the many possible cathode materials, not at all limited to NMC, LFP and related variants. Improvements to electrolytes are also in the offing with, of course, the desire to improve Li-ion battery safety as well as performance, using solid or semi-solid materials.
What all this means for battery analytics is that there is a tremendous amount of useful data from the battery development and manufacturing process, even before we take the first measurement of Voc (open-circuit Voltage). Our battery analytics platform can and often is configured to accommodate these datasets – let’s broady call them metrology – that are far upstream in the battery manufacturing process. This affords several benefits, including most importantly the ability to automate a multi-step process refinement and parameter check workflow.
Let’s take an example of NMC powder prior to cathode deposition. Upon receipt of the NMC, companies will typically perform quality control checks with an analytical instrument, for example an XRD analysis. A highly trained analyst will prepare the sample, take the measurement, save the data in a proprietary format, then refine the XRD spectrum, thereby deriving some parameters. This may include the presence of lithium carbonate impurities, looking at average crystal size. These parameters are then compared to an acceptable range prior to slurry preparation.
Some of the instruments whose data is captured: X-ray Diffraction, TGA, iCAP-OES & BET. Data produced from these tools can be used to automate a multi-step process refinement and parameter check workflow. The totality of this serialized data can later be referenced downstream after the battery is manufactured and in service in the field to extend useful life and improve performance.
After the slurry is prepared, another analyst might perform a PSD measurement, again with a proprietary format and with a refinement comparing the derived parameters with an acceptable range. This process repeats for all of the materials with a variety of analytical techniques, including XPS, TGA, pycnometry, iCAP-OES & BET.
An operator typically will save the metrology data on a shared network drive, putting the serial number of the relevant material in the file name. Our battery analytics system then grabs the file (being compatible with many vendor proprietary formats), normalizes the data, then refines it as necessary. For instance, if you need to differentiate a TGA and pull out transition temperatures by hand, that can be automated.
The metrology technique refinement process is, in fact, fully automated, and all of the data is stored and associated with the material regardless of where it originally resided. Most critically, this normalized data view is then associated with the batteries that are eventually made from the material. See our previous issue on capturing battery assembly data using a real-life example. As a result, it becomes possible to query and associate downstream data – including battery life cycle data – with parameters of the materials that went into that battery in a fully automated fashion.
By capturing metrology data from materials before a battery is even built, it becomes possible to query and associate downstream life cycle data with parameters of the materials that went into the battery, in a fully automated fashion.
The approach is not limited to metrology of the sort discussed here, as it is also used with images. For instance, some of our customers rely on image-heavy workflows in which electron and optical microscopy (and even some tomography) are used to analyze various battery constructs. Here again, automated refinement can be used. Perhaps an SEM image of a powder is used to interrogate the powder more thoroughly than possible with PSD. Here we can automate that image analysis to produce the relevant parameters. (If you are still using ImageJ or similar to accomplish, please reach out to me!). In some cases in R&D contexts, SEM cross sections of solid state batteries are analyzed using unsupervised machine learning to cluster good and bad battery topologies for some next-generation designs. In all of these cases, the ability to store and thread visual data is crucial.
SEM image of LTO powder with magnification: a-10.000X; b-20.000X (reference).
Our most mature customers extend this approach not only to automating the analysis of metrology data in order to understand their powders, but to automate the calibration and performance characterization of their metrology equipment. This allows them to automate these measurements as well, dramatically improving their reliability.
Battery material manufacturers already know that automating metrology refinement and threading (i.e. the association of data to serial number as assets are built up) aids in repeatability and scaling. This, in turn, has a material impact on service life, performance and quality of battery solutions developed from those materials.
For more insights into how threading serialized battery data can reduce risk and increase performance, please see our previous article: The importance of battery lot traceability.
