KAIST: "AI Minimizes Trial and Error in Battery Material Development"
An artificial intelligence (AI) technology capable of predicting battery performance has been developed. This technology is expected to minimize trial and error in the material development process by allowing AI to conduct numerous experiments required for battery material development and by increasing the reliability of predictions.
On January 26, KAIST announced that a research team led by Professor Seungbum Hong from the Department of Materials Science and Engineering, in collaboration with Professor Eun Ae Cho’s team, has developed an AI machine learning framework that can accurately predict the particle size of battery cathode materials and provide reliability information, even when experimental data is incomplete.
The cathode material inside a battery is a key component that enables lithium-ion batteries to store and release energy. Currently, the most widely used cathode materials in electric vehicle batteries are NCM-type metal oxides, which are mixtures of nickel (Ni), cobalt (Co), and manganese (Mn). These materials affect the battery’s lifespan, charging speed, driving range, and safety.
The joint research team focused on the fact that the size of the fine particles (primary particles) that make up the cathode material is a critical factor in determining battery performance. If the particles are too large, performance decreases, while if they are too small, stability issues may arise.
Traditionally, determining particle size required countless experiments, altering sintering temperature, time, and material composition. However, in real research settings, it is difficult to measure every condition without omission, and experimental data is often missing, making it challenging to precisely analyze the relationship between process conditions and particle size.
To address these challenges, the joint research team designed an AI framework capable of accurately predicting and controlling particle size. This framework was completed by combining a technology (MatImpute) that compensates for missing experimental data based on chemical properties and a probabilistic machine learning model (NGBoost) that calculates prediction uncertainty.
In particular, this model not only predicts particle size but also provides information on how reliable the prediction is, offering an important criterion for deciding under which conditions to synthesize materials in practice.
When the model was trained with expanded real experimental data, the AI demonstrated a high prediction accuracy of 86.6%. Analysis showed that the particle size of the cathode material is more influenced by process conditions such as baking temperature and time than by material composition, which aligns with existing experimental understanding.
Professor Seungbum Hong (front row left), Professor Eun Ae Cho (front row right), and other key researchers are gathered for a commemorative photo. Provided by KAIST
View original imageTo verify the reliability of the AI predictions, the joint research team synthesized four new cathode material samples with the same NCM811 composition (Ni 80%, Co 10%, Mn 10%), using manufacturing conditions not included in the existing data, and conducted experiments.
The results showed that the particle sizes predicted by the AI closely matched the actual measurements taken with a microscope, with most errors being less than 0.13 micrometers, which is much smaller than the thickness of a human hair.
Most importantly, the actual experimental results fell within the uncertainty range suggested by the AI, confirming the validity of both the predicted values and their reliability, according to the joint research team.
This research is significant in that it opens the way to identifying conditions with a high probability of success without having to conduct every possible experiment in battery research. The joint research team expects that this will accelerate the development of battery materials and significantly reduce unnecessary experiments and costs.
Professor Hong stated, "The core of this research is that AI simultaneously provides both the predicted value and the degree of trust in the prediction," adding, "We expect these results to be of practical help in designing next-generation battery materials more quickly and efficiently."
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
Meanwhile, Benediktus Madika, a doctoral researcher in the Department of Materials Science and Engineering, participated as the first author in this study. The research findings (paper) were recently published in the international journal 'Advanced Science,' which specializes in materials science and chemical engineering.
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.