Ionic liquids (ILs) are a class of molten salts with a collection of exciting properties, which have been employed for wide-ranging applications across chemistry, biology, and materials science. However, their inherently high viscosity hampers the ability of molecular dynamics (MD) simulations to explore their structure-property relationships on large spatiotemporal scales. Coarse-grained (CG) models address this challenge by retaining essential structural features while eliminating some atomic details, significantly reducing computational cost. A team of theoretical chemists presented the latest advances in IL CG models, with particular emphasis on the procedures for developing efficient CG models and parameterization methods with the aid of machine learning models. They also summarized applications of CG models in biological and electrochemical systems. This work is published in Industrial Chemistry & Materials on 09 Jun 2025.
“Constructing CG models tailored to specific ILs often involves determination of molecular representation and complex parameterization processes, especially for polarizable CG models, which can be a major hurdle for new practitioners entering this field. We anticipate that this review will provide a general understanding of foundational workflows, enabling researchers to design their own IL CG models. The team begins by summarizing parameterization methods, highlighting their respective advantages and limitations to assist readers in selecting the most suitable approach for their needs. Recently, machine learning (ML) techniques have garnered growing interest across various aspects of molecular simulations. In this review, we introduce their latest applications in IL CG modeling and share our insights on this evolving field”, explains Jing Ma, a professor at the Nanjing University.
The first step in constructing a CG model is to define the particles that represent the system. Compared to all-atom (AA) models, CG models require an additional AA-to-CG structural transformation, commonly referred to CG mapping. This transformation treats a group of atoms into a certain CG particle based on predefined rules. This team has introduced two strategies for determining CG mapping, complemented by literature examples, including direct methods based on chemical intuition and systematic methods, such as graph-based frameworks, relative entropy theory, and variational autoencoder techniques.
In the context of classical MD simulations, molecular interactions are primarily described by empirical potential energy functions, which determine the energies of configurations and the forces acting on each interaction site. Once the CG representation is determined, different parameterization strategies are available for CG potentials, including the “top-down” strategy, where CG models are directly fitted to a set of macroscopic experimental properties, and the “bottom-up” approach, where CG models utilize statistical mechanics principles to preserve the microscopic properties of the atomistic models. In this review, considerable efforts have been devoted to overviewing the theoretical foundations and numerical implementation details of various bottom-up CG methods.
The nature of ILs lies in their composition of ions, which generate local electric fields that polarize the surrounding molecular environment. The quality of the description of the electrostatic polarization effect in the FFs is crucial for the prediction of various properties. The team discusses two approaches for incorporating polarization effects into CG models: (a) the physics-inspired classic Drude oscillator model; and (b) quantum mechanics (QM)-based mean-field approach. The former approach introduces auxiliary Drude particles attached to the nuclei of core CG particles to reproduce inducible dipoles. Besides, many QM calculations on ion pairs or large-sized IL clusters indicate that polarization-induced charge transfer leads to a reduction in net ionic charge. By taking advantage of the fragment-based QM methods, such as the generalized energy-based fragmentation (GEBF) approach, polarization effects can be described at the QM level by using a large molecular cluster to mimic the bulk polarization environment. Two typical polarizable IL CG models developed by the two methods described above have been introduced with specific emphasis on the target data used for parametrization.
Thanks to advances in computing power and the abundance of high-quality data, ML techniques have been widely adopted for developing interatomic potentials in MD simulations. Their use primarily falls into two categories: constructing ML potentials (MLPs) and serving as optimization tools for determining parameters under the framework of classic FFs. The MLPs eliminate the empirical function forms by representing interatomic potentials as functions of descriptors that capture the local atomic environments. The second strategy employs ML as a surrogate model to assist in parameter optimization of classic FFs, aiming to enhance efficiency while preserving physical interpretability. The team offers an in-depth discussion of the current challenges associated with both approaches in developing CG models tailored for ILs.
The team finally introduces the applications of IL CG models, including condensed-phase, biological and electrochemical systems. The main obstacles to their applications in these systems lie in the transferability. The transferability refers to the ability of a CG model to maintain similar predictive accuracy across a wider range of thermodynamic state points (e.g., temperature, pressure, and composition) beyond the ones for parameterization. The team discusses several strategies for the development of more transferable and general IL CG models, such as the introduction of environment-dependent CG potentials, inclusion of more target data across multiple thermodynamic state points, and development of end-to-end automatic parametrization tools. Looking ahead, the research team hopes that future studies will build upon recent advances to address more complex systems for ILs. We aim for this review to serve as a valuable contribution to this exciting field.
The team of theoretical chemists includes Dr. Yang Ge, Dr. Qiang Zhu, Ms. Xueping Wang, and Prof. Jing Ma* from the Nanjing University; Dr. Qiang Zhu is currently affiliated with the University of Wollongong.
This work was supported by the National Natural Science Foundation of China (Grant No. 22033004, 22373049) and the Natural Science Foundation of Jiangsu Province (BK20232012).
Industrial Chemistry & Materials is a peer-reviewed interdisciplinary academic journal published by Royal Society of Chemistry (RSC) with APCs currently waived. ICM publishes significant innovative research and major technological breakthroughs in all aspects of industrial chemistry and materials, especially the important innovation of the low-carbon chemical industry, energy, and functional materials. Check out the latest ICM news on the blog.
END