FastTrack: Ion diffusivity calculation made easy

A research team from the Institute of Physics, Chinese Academy of Sciences, has developed FastTrack, a new machine learning-based framework dedicated  to evaluate ion migration barriers in crystalline solids. By combining machine learning force field (MLFFs) with three-dimensional potential energy surface (PES) sampling and interpolation, FastTrack enables accurate prediction of atomic migration barriers within mere minutes. Unlike traditional methods such as density functional theory (DFT) and nudged elastic band (NEB), which can take hours or days per calculation. FastTrack offers a speedup of over 100 times without sacrificing accuracy, closely matching experimental and quantum-mechanical benchmarks. This powerful tool automatically identifies diffusion pathways, visualizes energy landscapes, and provides detailed microscopic insights into ion migration mechanisms, crucial for designing more efficient batteries, fuel cells, and other energy storage and conversion devices.

The Obstruction

Ion diffusion, a process by which atoms and ions migrate within solids, is a fundamental phenomenon underlaying numerous natural and technological systems. In energy applications such as lithium-ion batteries, fuel cells, and solid-state electrolytes, ion transport significantly impacts device performance, longevity, and safety.. The migration barrier, or activation energy, determines how easily ions move through a solid and directly impacts material performance. Accurate characterization of atomic migration mechanisms and associated energy barriers is crucial for designing advanced materials with enhanced ionic conductivity and stability.

Computational methods, like density functional theory (DFT) combined with techniques like the nudged elastic band (NEB), have been crucial to study diffusion pathways. But, these approaches are computationally intensive, often requiring hours or days for a single pathway, which hampers their application in high-throughput material screening.. Ab initio molecular dynamics (AIMD) can also capture collective diffusion but is also computationally costly, while empirical models lack sufficient accuracy. In recent years, machine learning force fields (MLFFs) have emerged as a promising avenue to bridge this gap. By learning potentials directly from quantum mechanical data, MLFFs enable rapid, accurate predictions of atomic interactions in complex systems, greatly reducing computational costs. Despite these advances, efficiently integrating MLFFs with techniques that can systematically explore energy landscapes to identify diffusion pathways remains an ongoing challenge.

Tracking ion diffusion in a fast way

The researchers have introduced "FastTrack," a novel approach that bypasses the limitations of previous methods. Instead of relying on computationally expensive DFT calculations for every step, FastTrack leverages the power of MLFFs to rapidly generate a complete 3D PES for a migrating atom. By combining this with an efficient interpolation and pathfinding algorithm, the software can identify the minimum-energy pathway and calculate the migration barrier without needing pre-defined images, a requirement of traditional methods. To support this work, the team is also releasing FastTrack as an open-source software package, making it accessible to the global research community

More than a proof-of-concept

In layered LiCoO₂, FastTrack revealed distinct migration pathways: a ~600 meV barrier for single-vacancy diffusion versus ~250 meV for divacancy conditions, consistent with previous studies.

In LiFePO₄, the method captured the narrow one-dimensional channels along the [010] direction, yielding a ~300 meV barrier and highlighting the intrinsic rigidity of the phosphate framework.

Versatile design for various AI force fields

FastTrack is force-field agnostic, meaning it can be paired with any compatible MLFF. The team systematically evaluated three state-of-the-art models—GPTFF, CHGNet, and MACE—demonstrating their consistent performance across diverse chemistries. Task-specific fine-tuning with PBE/PBE+U datasets further sharpened barrier predictions, underscoring the importance of high-quality training data.

Removing blockages

For years, researchers used fast but imprecise tools like the bond valence method to search for ion-conducting materials, while accurate DFT-based approaches were too slow for large-scale screening. This trade-off has been the key bottleneck in battery materials discovery. FastTrack breaks through this barrier: it delivers near-DFT accuracy within minutes, enabling quantitative, high-throughput screening of ion transport across vast material spaces. With its open-source design, interactive visualization, and automated pathfinding, FastTrack turns a long-standing obstacle into a practical tool for accelerating energy materials research.

Software availability: github.com/atomly-materials-research-lab/FastTrack

Reference: Hanwen Kang, Tenglong Lu, Zhanbin Qi, Jiandong Guo, Sheng Meng and Miao Liu. FastTrack: a fast method to evaluate mass transport in solid leveraging universal machine learning interatomic potential[J]. AI for Science, 2025, 1(1). DOI: 10.1088/3050-287X/ae0808

---

---