Dendrites and dead lithium formation over prolonged cycling have long been challenges that hinder the safe implementation of metallic Li anodes. Herein, we employ polymer-stabilized liquid metal nanoparticles (LM-P NPs) of eutectic gallium indium (EGaIn) to create uniform Li nucleation sites enabling homogeneous lithium electrodeposition. Block copolymers of poly­(ethylene oxide) and poly­(acrylic acid) (PEO-b-PAA) were grafted onto the EGaIn surface, forming stabilized, well-dispersed NPs. Using a scalable spray coating approach, LM-P NPs were fabricated on copper current collectors, providing lithiophilic PEO sites and interactive carboxyl groups to guide Li deposition. The Li-EGaIn alloying process greatly reduced the Li+ diffusion barrier, enabling fast Li transport through the coating layer, resulting in decreased nucleation overpotential. Therefore, about five times lower Li nucleation overpotential was obtained on the LM-P modified Cu with an optimal composition of the polymers than the bare Cu substrates. DFT computations was used to reveal the binding properties between the LM-P layer and Li. Due to the regulated Li plating/stripping process, as-obtained 30 μm Li anodes paired with LiNi0.8Co0.1Mn0.1O2 (NCM811) with a negative/positive electrode capacity (N/P) ratio ∼ 10 exhibited stable cycling performance at 0.5C for over 250 cycles, with an average Coulombic efficiency of 99.55%. Ultrathin Li (1 μm) anodes with an N/P ratio ∼ 0.6 were also demonstrated in Li|LiFePO4 cells, which examined the stabilization of Li by LM-P NPs and monitored practical loadings of Li anodes that are close to anode-free systems.

No abstract

The lithium-metal anode is one of the most promising candidates for "beyond-lithium-ion" batteries thanks to its high specific capacity and low negative electrochemical potential. However, the electrode−electrolyte interface instability hinders its commercialization in rechargeable batteries. During cycles of charging and discharging, the lithium-metal anode is electrochemically plated and stripped along with the morphological evolution, which determines the cycling performance. In this work, with a phase-field model, we computationally characterize the morphological evolution dynamics during the plating and stripping steps at the lithium−metal−electrolyte interface. Our model is valid in a wide range of lithium concentrations in liquid electrolytes by incorporating nonidealities of electrolyte solutions into the interfacial reaction kinetics. Intriguingly, at fast stripping, i.e., high discharging overpotential, we observe an unexpected localized recrystallization phenomenon in high-lithium-ion-concentration valley regions. This recrystallization phenomenon mitigates the overall reaction rate heterogeneity and provides a potential approach to improving the morphological stability. Furthermore, we systematically investigate the correlation between the recrystallization phenomenon and lithium-ion activity and draw a simplified phase diagram for the overpotential-dependent recrystallization.

Rapid prediction of environmental chemistry properties is critical towards the green and sustainable development of chemical industry and drug discovery. Machine learning methods can be applied to learn the relations...

Rapid prediction of environmental chemistry properties is critical towards the green and sustainable development of chemical industry and drug discovery. Machine learning methods can be applied to learn the relations between chemical structures and their environmental impact. Graph machine learning, by learning the representations directly from molecular graphs, may enable better predictive power than conventional feature-based models. In this work, we leveraged graph neural networks to predict environmental chemistry properties of molecules. To systematically evaluate the model performance, we selected a representative list of datasets, ranging from solubility to reactivity, and compare directly to commonly used methods. We found that the graph model achieved near state-of-the-art accuracy for all tasks and, for several, improved the accuracy by a large margin over conventional models that rely on human-designed chemical features. This demonstrates that graph machine learning can be a powerful tool to do representation learning for environmental chemistry. Further, we compared the data efficiency of conventional feature-based models and graph neural networks, providing guidance for model selection dependent on the size of datasets and feature requirements.