Fabrication of High-Quality Thin Solid-State Electrolyte Films Assisted by Machine Learning

Abstract

circumvent flammability concerns of liquid electrolytes. However, enhancing energy densities by thinning SSE layers and enabling scalable coating processes remain challenging. While previous studies have addressed thin and flexible SSEs, mainly ionic conductivity was considered for performance evaluation, and no systematic research on the effects of manufacturing conditions on the quality of SSE films was performed. Here, both uniformity and ionic conductivity are considered for evaluating the SSE films under the guidance of machine learning (ML). Three algorithms, principal component analysis, K-means clustering, and support vector machine, are employed to decipher the interdependencies between manufacturing conditions and film performance. Guided by ML, a 40 μm SSE film with high ionic conductivity and good uniformity is used to construct a LiNi 0.8 Co 0.1 Mn 0.1 O 2 || Li 6 PS 5 Cl || LiIn cell demonstrating 100 cycles. This study presents an efficient ML-assisted approach to optimize scalable production of high-quality SSE films.

A Versatile Li6.5In0.25P0.75S5I Sulfide Electrolyte Triggered by Ultimate‐Energy Mechanical Alloying for All‐Solid‐State Lithium Metal Batteries

“…Currently, with the popularity of electric cars and portable electronic devices, lithium‐ion batteries (LIBs) have been generally utilized in daily life. [ 1,2 ] However, the use of combustible organic electrolytes in LIBs has raised significant concerns about the safety. [ 3,4 ] In addition, the limited energy density further hinders the extensive application of LIBs.…”

Smart Materials Prediction: Applying Machine Learning to Lithium Solid-State Electrolyte

“…All the algorithms are not independent of each other. The data can be analyzed by comprehensive statistical tests of several algorithms in algorithm modeling [ 78 ] to obtain the best results. We can note that in many of the above algorithms, neural networks can learn layer by layer on the input and produce high learning rates, so neural network algorithms are often combined with other algorithms to build prediction models and thus obtain higher accuracy on the data results.…”

Perspectives on manufacturing simulations of Li-S battery cathodes

“…After this, many other researchers have utilized ML methods to inform experimental designs for LIBs. Chen et al [124], in collaboration with our group, used similar methods to decipher parameter interdependencies on the fabrication of highly conductive and homogeneous solid-state electrolyte films. Similarly, Duquesnoy et al [125] from our group used the supervised Gaussian Naives Bayes (GNB) technique in order to assess the probability of casting homogeneous LIB electrodes depending on AM percentages, comma gaps, and slurry solid contents.…”