A domestic industry-academia cooperation research team has established an 'autonomous exploration laboratory' that develops secondary battery cathode materials without researcher intervention, reducing the exploration period by 93%.
Professor Seo Dong-hwa and his research team at the Korea Advanced Institute of Science and Technology (KAIST) announced on the 3rd that they have built an autonomous exploration laboratory for exploring secondary battery cathode materials based on artificial intelligence (AI) and automation technology in cooperation with the energy materials research institute of POSCO Holdings.
Secondary battery cathode materials must meet challenging criteria, including high charge rates, energy density, and stability, which necessitates considering numerous candidate materials for development. This needs the high labor input of skilled researchers and lengthy development times.
The research team established an autonomous exploration laboratory based on an automation system that performs sample quantification, mixing, pelletizing, sintering, and analysis without researcher intervention, along with an AI model that interprets the analyzed data and learns from it to select the best candidates. This was achieved by constructing each process as an individual device module, which is handled by a central robotic arm to increase experimental efficiency.
The research team also introduced a high-speed sintering method different from traditional methods to improve synthesis speed. Sintering is the process of heating samples to consolidate powder particles into a single mass. As a result, the time required for the sintering process was reduced by 50 times, and 12 times more material data could be obtained compared to existing researcher-based experiments.
The secured material data is automatically interpreted through the AI model to extract impurity ratios and other information. This data is systematically stored, regardless of synthesis success, to build a high-quality databases, and a system has been implemented that utilizes this data as learning data for optimization AI models.
When the intelligent experimental automation system operates 24 hours, it can provide about 12 times more experimental data, and the material exploration time can be reduced by 93%. Assuming that 500 experiments are needed for material exploration, the traditional method, where researchers perform the experiments directly, would take 84 days, but the automation system can complete it in about 6 days.
Professor Seo Dong-hwa noted that this is a technology to solve the reduction of research personnel due to low birth rates and explained that he expects to accelerate secondary battery material development and enhance global competitiveness based on high-quality material data.
POSCO Holdings' future technology research institute aims to dramatically increase the speed of next-generation secondary battery material development by applying an upgraded version of the autonomous exploration laboratory system to its own research laboratory after 2026.