AI Materials doi: 10.3390/aimater1020006

Authors: Shucheng Xing Jian Zhou Zhimei Sun

Zentropy theory has emerged as a multiscale thermodynamic framework that bridges quantum mechanics, statistical mechanics, and macroscopic materials behavior by embedding internal degrees of freedom within configurational ensembles. This review summarizes its theoretical foundations, representative applications, current limitations, and future directions. By incorporating intrinsic configurational entropy and free-energy-based statistical weighting, zentropy theory enables improved descriptions of phase stability, thermal expansion, and phase transitions in materials such as ferroelectrics, magnetic systems, high-entropy materials, and superconductors. Recent extensions also connect zentropy with artificial intelligence through data-driven thermodynamic modeling. Despite these advances, several challenges remain, including the ambiguity of configurational coarse-graining, strong cross-degree-of-freedom coupling, propagation of density functional theory errors, and limited applicability to delocalized or non-crystalline states. Future progress will require theoretical advances, including non-ergodic extensions, rigorous mathematical treatment of recursive multiscale entropy, and improved descriptions of low-temperature quantum effects. These efforts should be complemented by standardized software workflows, machine learning integration, and robust uncertainty quantification. Addressing these bottlenecks will help to further develop zentropy theory as a critically assessed framework for multiscale thermodynamic modeling and materials design.

AI Materials doi: 10.3390/aimater1020005

Authors: Guangyao Xu Xiaolong Wei Changhui Shen Tongtong Song Hongchen Chu Jie Luo Yun Lai

Artificial intelligence (AI), particularly deep learning (DL), is revolutionizing optical metamaterial design by overcoming the fundamental challenges of multidimensional parameter spaces, nonlinear structure–property relationships, and the intrinsic non-uniqueness of inverse problems. By learning complex mappings between geometric structures and electromagnetic responses, DL enables rapid forward prediction and on-demand inverse design without computationally intensive full-wave simulations. This review provides a comprehensive survey of AI-driven design methodologies across four key metamaterial platforms: localized resonant nanostructures, metasurfaces, periodic and guided-wave photonic structures, and complex scattering systems. For each platform, we systematically examine the neural network architectures employed, the specific design challenges addressed, and the representative achievements attained. These data-driven approaches not only significantly accelerate the discovery of high-performance structures but also offer new opportunities for extracting physical insights into light–matter interactions. We assess the critical challenges of data efficiency, model interpretability, and experimental feasibility, and outline emerging research directions that may address these barriers. This review aims to provide both a comprehensive summary of the current state of the art and forward-looking perspectives for this rapidly evolving interdisciplinary field.

AI Materials doi: 10.3390/aimater1010004

Authors: Shengxiang Liu Jingfeng Wang Zihe Li Wenyan Jiao Fuyun Lv Huijun Liu

Due to their unique structures, intriguing electronic properties, and potential applications across various fields, Janus materials have attracted extensive attention from the science community. However, the thermal transport properties of Janus systems are less known so far, especially regarding lattice thermal conductivity (LTC). In this work, we establish an accurate machine learning potential by which the phonon Boltzmann transport equation can be iteratively solved to readily predict the LTC of Janus WXY (X, Y = S, Se, Te) monolayers. It is found that the LTC for all three systems decreases monotonically with increasing temperature. Among them, the WTeSe monolayer exhibits the lowest LTC, which can be traced back to the competition between the contributions of phonon group velocity and relaxation time. Interestingly, we demonstrate that the effect of four phonon scattering plays an important role in accurately determining the LTC of these Janus monolayers. Our work also provides an alternative way of effectively predicting the LTC of systems with low symmetry and/or large size.

AI Materials doi: 10.3390/aimater1010003

Authors: Rubén Guerrero-Rivera Norma A. García-Vidaña Francisco J. Godínez-García Zhipeng Wang Morinobu Endo Josué Ortiz-Medina

The representation of electronic wavefunctions in real space grids, which are directly related to molecular orbitals and electronic densities either in molecular or crystalline systems, is a fundamental part of many studies at ab initio levels, since it contributes to the understanding of complex physical and chemical phenomena at the nanoscale. This work proposes the use of a deep convolutional neural network for the prediction of electronic wavefunctions at arbitrary positions along high-symmetry points within the reciprocal space (first Brillouin zone), which can be represented as isosurfaces in the real space. The proposed neural network algorithm is trained with data from density functional theory (DFT) calculations of monolayer 2D crystalline systems (i.e., pristine, B- and N-doped graphene, and MoS2) and was able to produce predictions of data for wavefunction representation on the real space, with accuracies in between 62% and 92%, from calculated determination coefficients. Moreover, the optimized method for generating spatial representations of electronic wavefunctions, based on Machine Learning, is at least 25× faster than the conventional DFT-based methodology, enabling an efficient way for a quick assessment of 2D material properties related to the spatial distribution of electronic wavefunctions in the real space, such as local charge density and molecular orbital visualization in crystalline systems, and including their dependence on the position within the reciprocal space.

AI Materials doi: 10.3390/aimater1010002

Authors: Kingsley Yeboah Gyabaah Bernard Mahoney Anthony Kwasi Martey Cheng Yan Patrick Mensah Guoqiang Li

Additive manufacturing (AM) of polymers and polymer composites is changing how customized, lightweight, and complex parts are produced across various industries. However, predicting the final properties of printed parts remains challenging due to variations in material compositions, processing conditions, and microstructural characteristics. This review explores how machine learning (ML) is being used to address these challenges. It examines the application of various ML approaches in polymer and polymer composite design for AM, including supervised, unsupervised, semi-supervised, self-supervised, and reinforcement learning, for predicting key properties such as mechanical strength, thermal stability, and electrical performance. The review also highlights hybrid techniques that combine ML with physics-informed modeling, including the use of digital twins, to enhance AM process control. Challenges and future perspectives, such as data scarcity, model interpretability, and computational demands, are discussed. In summary, ML is showing strong potential to support faster, more reliable, and more sustainable development of advanced polymers and polymer composites for AM.

AI Materials doi: 10.3390/aimater1010001

Authors: Xin-Gao Gong

We stand at a transformative moment in scientific history, where artificial intelligence and materials science are engaged in a mutually reinforcing cycle of advancement [...]