Saturday, November 22, 2025

Low-temperature electrolytes for lithium-ion batteries: Current challenges, development, and perspectives




Shanghai Jiao Tong University Journal Center
Low-Temperature Electrolytes for Lithium-Ion Batteries: Current Challenges, Development, and Perspectives 

image: 

  • Key electrolyte-related factors limiting the low-temperature performance of lithium-ion batteries (LIBs) are analyzed.
  • Emerging strategies to enhance the low-temperature performance of LIBs are summarized from the perspectives of electrolyte engineering and artificial intelligence (AI) -assisted design.
  • Perspectives and challenges on AI-driven design, advanced characterization, and novel electrolyte systems for low-temperature LIBs.
view more 

Credit: Yang Zhao, Limin Geng*, Weijia Meng*, Jiaye Ye*.





As electric vehicles, satellites and wearable electronics push into sub-zero environments, conventional lithium-ion batteries (LIBs) lose most of their energy and power, while lithium plating threatens safety. Now, researchers from Chang’an University and Queensland University of Technology, led by Professor Limin Geng, Professor Weijia Meng and Dr Jiaye Ye, have published a forward-looking review on low-temperature (LT) electrolytes that keep LIBs charging and discharging down to −80 °C. This work offers a systematic roadmap for next-generation energy-storage systems that thrive in the cold.

Why LT Electrolytes Matter
   • Energy Efficiency: Rational molecular design slashes Li⁺ desolvation energy to <40 kJ mol-1, cutting polarization losses and eliminating external heaters that drain 10–20 % of pack energy.
   • In-Battery Transport: High-entropy and weakly-solvating formulations maintain ionic conductivity >1 mS cm-1 at −60 °C, enabling 4 C charge rates without lithium plating.
   • Extreme-Weather Applications: From Mars rovers to Arctic drones, LT electrolytes unlock reliable power where traditional LIBs cease to function.

Innovative Design and Features
   • Electrolyte Families: The review covers ester-based (methyl acetate, ethyl difluoroacetate), ether-based (DOL/DME, THF, CPME), nitrile-based (fluoroacetonitrile) and gel-polymer systems, detailing how freezing point, dielectric constant and donor number dictate Li⁺ solvation structure.
   • Functional Components: Dual-salt (LiFSI-LiDFOB), ternary-anion (LiPF6-LiTFSI-LiNO3) and AI-screened additives (LiTDI, NaPFO) are highlighted for building LiF- or Li3PO4-rich SEI/CEI layers that reduce interfacial resistance ten-fold.
   • AI-Guided Formulation: Machine-learning models trained on >150 000 molecular candidates predict melting point, viscosity and LUMO energy within 5 K or 0.1 eV, accelerating electrolyte discovery from months to hours.

Applications and Future Outlook
   • Multi-Level Screening: High-throughput DFT plus SHAP interpretability identifies dipole moment and molecular radius as key descriptors, delivering non-fluorinated ethers that cycle 300 times at −30 °C with 99 % capacity retention.
   • Digital Logic Gates: LT gel-polymer electrolytes enable flexible printed-circuit modules that operate at −40 °C, providing a new route for cold-weather in-memory computing and IoT sensors.
   • Artificial Interphases: From self-assembled NaPFO monolayers to organosilicon-rich SEIs, these nano-films suppress dendrites and raise Coulombic efficiency to 97.5 % at −60 °C.
   • Challenges and Opportunities: The review pinpoints the need for standardized LT testing protocols, physics-informed neural networks that couple solvation structure to plating propensity, and automated robotic platforms that translate AI predictions into litre-scale synthesis. Future work will target high-entropy electrolytes, phase-diagram-guided formulations and cryogenic in-situ NMR to close the gap between lab demos and commercial 8 Ah pouch cells.

   This comprehensive roadmap provides materials scientists, cell engineers and AI researchers with a common language for co-optimizing salts, solvents, additives and polymers for sub-zero operation. Stay tuned for more breakthroughs from Professor Limin Geng, Professor Weijia Meng and Dr Jiaye Ye!

 

Deep learning-assisted organogel pressure sensor for alphabet recognition and bio-mechanical motion monitoring



Shanghai Jiao Tong University Journal Center
Deep Learning-Assisted Organogel Pressure Sensor for Alphabet Recognition and Bio-Mechanical Motion Monitoring 

image: 

  • We rationally designed a robust, biocompatible CoN CNT/PVA/GLE organogel with self-healing, anti-freezing, and self-adhesive properties for wearable sensing applications.
  • Incorporation of CoN CNT enables high-performance, stable pressure sensing for up to one month, with a sensitivity of S = 5.75 kPa-1, r2 = 0.978 in the detection range 0-20 kPa, with robust operation under high humidity and extreme temperatures (−20 to 45 °C).
  • It accurately detects English alphabets, achieving 98% recognition accuracy using deep learning models.
view more 

Credit: Kusum Sharma, Kousik Bhunia, Subhajit Chatterjee, Muthukumar Perumalsamy, Anandhan Ayyappan Saj, Theophilus Bhatti, Yung-Cheol Byun, Sang-Jae Kim*.




As wearable electronics migrate toward real-time health monitoring and seamless human–machine interfaces, conventional hydrogels freeze, dry out and fracture under daily conditions. Now, a multidisciplinary team led by Prof. Sang-Jae Kim (Jeju National University) has unveiled a CoN-CNT/PVA/GLE organogel sensor that marries sub-zero toughness with AI-grade pattern recognition. The device delivers 5.75 kPa-1 sensitivity across 0–20 kPa, heals in 0.24 s, and classifies handwritten English letters at 98 % accuracy—offering a robust, bio-compatible platform for next-generation soft robotics and personalized healthcare.

Why the CoN-CNT Organogel Matters
   • Freeze-Tolerant & Anti-Dehydration: Binary ethylene-glycol/water solvent and Co–Nx coordination keep conductivity at 1.10 mS cm-1 down to −20 °C and 95 % RH for >75 days.
   • Self-Healing & Adhesive: Dynamic borate-ester bridges and hydrogen bonding restore 88 % mechanical strength in 60 min and stick stably to skin, wood, glass and curved plastics.
   • AI-Ready Sensing: Piezo-capacitive response captures stroke pressure, lift-off and curvature, enabling 1D-CNN + XGBoost models to discriminate all 26 letters and digits with <2 % error.

Innovative Design and Features
   • Hybrid Conductive Network: Cobalt-nanoparticle@nitrogen-doped CNTs provide metallic pathways, interfacial polarization and antioxidant shells, outperforming pristine CNT or ionic fillers.
   • Dual-Crosslink Matrix: FDA-recognized PVA and biodegradable gelatin form reversible boronate esters; EG plasticizer suppresses ice crystallization and maintains chain mobility.
   • Deep-Learning Pipeline: Sliding-window feature extraction → CNN-LSTM temporal encoder → XGBoost meta-classifier; robust to variable writing speed and pressure (95 % accuracy under perturbation).

Applications and Future Outlook
   • Multimodal Health Patches: Real-time tracking of finger/wrist bending, throat vibrations during speech and gait asymmetry for rehabilitation and tele-medicine.
   • Soft Robotics Interface: Ultra-low detection limit (≈20 Pa) enables tactile feedback for prosthetic grasping and collaborative robot arms.
   • Challenges & Opportunities: Scaling roll-to-roll slot-die coating, integrating wireless BLE SoCs and extending vocabulary to Chinese characters and sign-language gestures are next milestones.

This work provides a comprehensive material-plus-AI blueprint for durable, intelligent wearable sensors that operate reliably from Arctic drones to tropical wearables. Stay tuned for further breakthroughs from Prof. Kim’s team!