Luo Haoran, Gou Qianzhi, Zheng Yujie, Wang Kaixin, Yuan Ruduan, Zhang Sida, Fang Wei, Luogu Ziga, Hu Yuzhi, Mei Huaping, Song Bingye, Sun Kuan, Wang John, Li Meng
National Innovation Center for Industry-Education Integration of Energy Storage, MOE Key Laboratory of Low-Grade Energy Utilization Technologies and Systems, CQU-NUS Renewable Energy Materials & Devices Joint Laboratory, College of Energy & Power Engineering, Chongqing University, Chongqing 400044, China.
School of Building Services Science and Engineering, Xi'an University of Architecture and Technology, Xi'an 710055, China.
ACS Nano. 2025 Jan 21;19(2):2427-2443. doi: 10.1021/acsnano.4c13312. Epub 2025 Jan 7.
The utilization of electrolyte additives has been regarded as an efficient strategy to construct dendrite-free aqueous zinc-ion batteries (AZIBs). However, the blurry screening criteria and time-consuming experimental tests inevitably restrict the application prospect of the electrolyte additive strategy. With the rise of artificial intelligence technology, machine learning (ML) provides an avenue to promote upgrading of energy storage devices. Herein, we proposed an intriguing ML-assisted method to accelerate the development efficiency of electrolyte additives on dendrite-free AZIBs. Concretely, we selected the Gutmann donor number (DN value) as a screen parameter, which can reflect the interaction between solvent molecules and ions, and proposed an integrated ML model that can predict the DN values of organic molecules via molecular fingerprints, thereby achieving the screening of electrolyte additives. Then, combined with experimental tests and theoretical calculations, the influence law of three additive molecules with different DN values on the thermodynamic stability of the Zn anode and its corresponding optimization mechanisms were revealed; the DN values of the additives are in positive correlation with the electrochemical performance of the Zn anode. Especially, an isopropyl alcohol (IPA) additive with a high DN value (36) integrated with various Zn-based cells presented a superior electrochemical performance, including a high calendar life (1500 h), a stable Coulombic efficiency (99% within 450 cycles), and a favorable cycling retention. This work pioneers ML techniques for predicting DN values for electrolyte additives, offering a compelling investigation method for the investigation of AZIBs.
使用电解质添加剂被认为是构建无枝晶水系锌离子电池（AZIBs）的一种有效策略。然而，模糊的筛选标准和耗时的实验测试不可避免地限制了电解质添加剂策略的应用前景。随着人工智能技术的兴起，机器学习（ML）为促进储能设备的升级提供了一条途径。在此，我们提出了一种有趣的ML辅助方法，以提高无枝晶AZIBs电解质添加剂的开发效率。具体而言，我们选择古特曼给体数（DN值）作为筛选参数，其可以反映溶剂分子与离子之间的相互作用，并提出了一种集成ML模型，该模型可以通过分子指纹预测有机分子的DN值，从而实现电解质添加剂的筛选。然后，结合实验测试和理论计算，揭示了三种具有不同DN值的添加剂分子对锌阳极热力学稳定性的影响规律及其相应的优化机制；添加剂的DN值与锌阳极的电化学性能呈正相关。特别是，具有高DN值（36）的异丙醇（IPA）添加剂与各种锌基电池集成后表现出优异的电化学性能，包括高日历寿命（1500小时）、稳定的库仑效率（在450次循环内为99%）和良好的循环保持率。这项工作开创了利用ML技术预测电解质添加剂DN值的先河，为研究AZIBs提供了一种引人注目的研究方法。
