Elucidating the impact of metal doping in Li1.15(Ni0.35Mn0.65)0.85O2 cathodes using high-throughput experiments and machine learning

Abstract

With the ever-increasing demand for Li-ion batteries amplifying the economic, environmental, and geopolitical issues of cobalt-containing electrodes, Li-rich Mn-based layered oxides (LMRs) are regarded as promising next generation cathode materials. Due to the projected lower costs and higher energy density than current-generation LiNi_xMn_yCo_zO₂ materials, LMRs have been the subject of extensive research. However, cycling stability remains a key challenge with transition metal (TM) dissolution during cycling being at the root of this problem. Several methods have previously been applied to reduce transition metal dissolution, including doping. However, there has been no systematic study of the impact of a wide variety of dopants on LMR materials, limiting our ability to gather foundational knowledge about the role of doping in these structures and predict their impact to efficiently screen promising materials. In this work, we applied high-throughput techniques established by the McCalla group to study both the structural and electrochemical characteristics of cathodes made at 3 different temperatures with 57 different dopants. For the first time, transition metal dissolution was studied in high-throughput through the elemental analysis of the Li anode after a high-temperature voltage hold. After collecting all these data, we employed machine-learning techniques to establish predictive links between various properties (both structural and electrochemical) and TM dissolution. Ultimately, we discovered 45 different dopant/temperature combinations that showed both improved specific discharge capacity and reduced TM dissolution such that valuable lessons were learned about the impact of dopants in these materials to facilitate further accelerated material design.