The efficient and economical exploitation of polymers with high thermal conductivity (TC) is essential to solve the issue of heat dissipation in organic devices. Currently, the experimental preparation of functional polymers with high TC remains a trial-and-error process due to the multi-degrees of freedom during the synthesis and characterization process. Polymer informatics equips machine learning (ML) as a powerful engine for the efficient design of polymers with desired properties. However, available polymer TC databases are rare, and establishing appropriate polymer representation is still challenging. In this work, we propose a high-throughput screening framework for polymer chains with high TC via interpretable ML and physical feature engineering. The hierarchical down-selection process stepwise optimizes the 320 initial physical descriptors to the final 20 dimensions and then assists the ML models to achieve a prediction accuracyR²over 0.80, which is superior to traditional graph descriptors. Further, we analyze the contribution of the individual descriptors to TC and derive the explicit equation for TC prediction using symbolic regression. The high TC polymer structures are mostly π-conjugated, whose overlapping p-orbitals enable easy maintenance of strong chain stiffness and large group velocities. Ultimately, we establish the connections between the individual chains and the amorphous state of polymers. Polymer chains with high TC have strong intra-chain interactions, and their corresponding amorphous systems are favorable for obtaining a large radius of gyration and causing enhanced thermal transport. The proposed data-driven framework should facilitate the theoretical and experimental design of polymers with desirable properties.

The integration of artificial intelligence (AI) into materials science has catalyzed a transformative revolution in energy storage technology, particularly in the development of advanced rechargeable battery systems. This paradigm shift is redefining traditional approaches to battery materials innovation by the emergence of AI‐driven methodology. The review commences with an overview of typical algorithms and workflows integrated in the design and optimization of rechargeable battery materials in recent years. Subsequently, the cutting‐edge applications of AI in the development of anode, cathode, liquid electrolyte, and solid‐state electrolyte materials are reviewed. The key performance metrics and application characteristics are summarized, and the most recent and innovative milestones are highlighted, emphasizing the ability of the AI‐driven method to solve complex multi‐parameter coupling relationships. Meanwhile, this paper briefly discusses the critical challenges impeding the full realization of AI's potential in battery innovation, including data scarcity, data quality, and model interpretability. Finally, the review outlines future directions for AI‐powered closed‐loop autonomous materials discovery systems, proposing a visionary framework that integrates high‐throughput experimental and computational platforms, standardized databases, physics‐informed algorithms, and explainable AI protocols. This synthesis of cross‐disciplinary expertise positions AI not just as an optimization tool but as a paradigm‐shifting force in the energy storage field.

Modern generative models based on deep learning have made it possible to design millions of hypothetical materials. To screen these candidate materials and identify promising new materials, we need fast and accurate models to predict material properties. Graphical neural networks (GNNs) have become a current research focus due to their ability to directly act on the graphical representation of molecules and materials, enabling comprehensive capture of important information and showing excellent performance in predicting material properties. Nevertheless, GNNs still face several key problems in practical applications: First, although existing nested graph network strategies increase critical structural information such as bond angles, they significantly increase the number of trainable parameters in the model, resulting in a increase in training costs; Second, extending GNN models to broader domains such as molecules, crystalline materials, and catalysis, as well as adapting to small data sets, remains a challenge. Finally, the scalability of GNN models is limited by the over-smoothing problem. To address these issues, we propose the DenseGNN model, which combines Dense Connectivity Network (DCN), hierarchical node-edge-graph residual networks (HRN), and Local Structure Order Parameters Embedding (LOPE) strategies to create a universal, scalable, and efficient GNN model. We have achieved state-of-the-art performance (SOAT) on several datasets, including JARVIS-DFT, Materials Project, QM9, Lipop, FreeSolv, ESOL, and OC22, demonstrating the generality and scalability of our approach. By merging DCN and LOPE strategies into GNN models in computing, crystal materials, and molecules, we have improved the performance of models such as GIN, Schnet, and Hamnet on materials datasets such as Matbench. The LOPE strategy optimizes the embedding representation of atoms and allows our model to train efficiently with a minimal level of edge connections. This substantially reduces computational costs and shortens the time required to train large GNNs while maintaining accuracy. Our technique not only supports building deeper GNNs and avoids performance penalties experienced by other models, but is also applicable to a variety of applications that require large deep learning models. Furthermore, our study demonstrates that by using structural embeddings from pre-trained models, our model not only outperforms other GNNs in distinguishing crystal structures but also approaches the standard X-ray diffraction (XRD) method.

The release of ChatGPT has triggered global attention on artificial intelligence (AI), and AI for science is thus becoming a hot topic in the scientific community. When we think about unleashing the power of AI to accelerate scientific research, the question coming to our mind first is whether there is a continuous supply of highly available data at a sufficiently large scale.

The informatics algorithm-driven approach overcomes the high-cost and time-consuming drawbacks of conventional trial-and-error procedures and enables efficient exploration of polymers with desired properties. Intrinsically thermally conductive polymers are highly in demand for effective thermal management of energy devices, flexible electronics and power modules, from the perspective of heat dissipation, operation reliability, and service lifetime of the devices. However, since most polymers are thermally insulating, the exploration of intrinsically thermally conductive polymers and understanding the relationship between molecular structures and thermal conduction are essential yet challenging tasks. In this work, we proposed and developed a hybrid framework that combines automated physical feature engineering and symbolic regression for exploring amorphous polymers with desired thermal conductivity. The feature engineering approach constructed 325 initial physical descriptors based on information from atoms, electrons, and charges in monomers, together with bond and non-bond interactions in amorphous systems. 25 optimized descriptors are finally figured out through statistical coefficient filtering and machine learning model selection. The relationship between the macromolecular structure and thermal conductivity was established with high accuracy, which outperforms other molecular representations of Morgan fingerprints, molecular access systems, and the dimensionality reduction technique of principal component analysis. Machine learning models based on optimized descriptors recommended 104 polymer candidates with good thermal conductivity out of over 790 000 polymers, of which 54.8% have thermal conductivities above 0.35 W m⁻¹ K⁻¹, significantly higher than those found in the training dataset. Furthermore, the feature importance ranking coupled with the explicit analytical models fitted by symbolic regression suggests that the conjugated molecular structure, chain stiffness, and maximum relative atomic mass in polymers have a strong influence on thermal conductivity. Our proposed approach provides a valuable tool for effectively exploring the relationship between macromolecular structures and thermal properties.