Revolutionizing materials design through advanced artificial intelligence-assisted multiscale simulation

Abstract

Materials science is currently at the forefront of technological development, which enables remarkable advancements in various aspects of our society. For the past few decades, multiscale computational simulations ranging from accurate first principles calculation and atomistic molecular dynamics to mesoscopic and macroscopic continuum models have been essential tools for understanding, predicting, and ultimately designing materials with desired properties.

Today, multiscale modeling/simulation is entering an exciting era where recent rapid advances in artificial intelligence (AI)/data-driven approach are beginning to converge with well-established multiscale computational simulation toolset. This convergence opens unprecedented opportunities not only for accelerating high-throughput screening of vast compositional and structural space across a wide variety of length/time scales but also for discovering new structure–property relationships for efficient designing and synthesizing materials for practical applications.

There is no doubt that AI/data-driven approaches are a central pillar within today's materials genome engineering paradigm for rapid materials discovery and design, which inspires us with honor and immense pleasure to bring forth the special issue of Materials Genome Engineering Advances (MGE Advances) entitled “Revolutionizing Materials Discovery by Advanced AI-Assisted Multiscale Computational Modeling” to realize its mission of breaking the barrier between disciplines and fostering digital, smart materials R&D. The following thematic selection highlights the active state-of-the-art advances in applications involving advanced machine learning/AI algorithms applied to a variety of multiscale computational modeling/numerical simulation efforts.

Materials discovery accelerated with high-throughput computing: For example, benefiting from high-throughput density functional theory calculations to search for candidates of antiperovskites possessing desired properties (such as large spin Hall conductivity) or guiding the design of particular alloy systems (for instance, easily separable Fe-containing intermetallics in Al–Si alloy).

The development and application of advanced ML potentials: Developing deep learning- or machine learning-based potentials enables atomistic simulations for complicated events at larger length/time scales or in harsh environments (e.g., investigating finite-temperature behaviors of materials such as NbO₂ or deciphering high-temperature deforming mechanisms of intermetallics such as Ni₃Al).

Data-aided design of novel materials: Adopting a data-driven strategy based on existing simulation or experimental datasets to design novel materials with desired functions (e.g., optimizing mechanical properties of biodegradable deformed zinc alloys).

Developing new computational methods/tools for science innovations: Improving existing computational infrastructure (e.g., developing new powerful software packages for molecular dynamics simulations that can be efficiently run on GPUs using popular open-source languages such as Python) allows more people to enjoy the significant simulation power. Benefiting from recent AI technologies beyond typical regression/classification ones (e.g., knowledge graph attention networks) for some special problems (e.g., intelligent manufacturing) and trying their best to improve their interpretability/accuracy.

Totally, there are 11 contributions of top-notch researchers from all over the world included in this collection, which demonstrates real progress toward “materials design through advanced AI-assisted multiscale simulation.” Thank you very much for sharing your new findings and insightful visions with us. In addition, we would also like to extend our sincere gratitude to the hardworking reviewers for their precious time and valuable feedback.

The combination of AI and multiscale simulations is undoubtedly pushing the limits of computation-based material design. We hope that this special issue will serve as a valuable resource for researchers, students, and practitioners in materials science, physics, chemistry, computer science, engineering, etc. We believe that the articles in this collection will not only provide a snapshot of the state-of-the-art discovery but also inspire further innovative research, promote new interdisciplinary collaborations, and make a significant contribution to accelerating the discovery and synthesization of novel materials.

Please take your time to enjoy the intelligent contributions in this special issue and be part of the ongoing materials design revolution.

Hui Wang: Writing—review and editing. Xing-Qiu Chen: Writing—Review and editing.

The authors declare no conflicts of interest.