Massive Discovery of 9,139 Low-Dimensional Materials from Materials Project via Universal Computational Strategy, Including 887 Exfoliable 2D Materials

August 9, 2026

ACS Chemistry of Materials USA

Overview

This research combined universal machine learning interatomic potentials (UMLIPs) with an advanced force constant (FC)-based dimensionality classification method to massively discover new low-dimensional materials from the Materials Project database. Through phonon calculations for 35,689 materials, 9,139 low-dimensional materials were identified (1,838 0D clusters, 1,760 1D chains, 3,057 2D sheets/layers, and 2,484 mixed-dimensional materials), with 887 of the 2D materials shown to be easily or potentially exfoliable. This demonstrates the power of computational materials science to accelerate innovative material discovery.

In Depth

Key Findings

This study leveraged universal machine learning interatomic potentials (UMLIPs) and an innovative force constant (FC)-based dimensionality classification method to discover a massive number of new low-dimensional materials from the Materials Project database. This computational strategy identified an unprecedented 9,139 diverse low-dimensional materials, including 887 exfoliable 2D materials, demonstrating its potential to dramatically accelerate the pace of discovery in materials science.

Technical / Clinical Details

The research team first developed highly versatile UMLIPs capable of covering a broad chemical space. UMLIPs can accommodate various elements and bonding states, enabling materials exploration on a far grander scale than traditional potentials tailored for specific elemental systems. Next, these UMLIPs were used to perform phonon calculations for 35,689 materials present in the Materials Project database. Phonon calculations are a key technique for elucidating the lattice vibrational properties of materials, essential for evaluating their stability and dimensionality. Based on the calculated phonon band structures and interatomic force constants (FCs), a new dimensionality classification algorithm was developed to automatically identify whether each material belonged to 0D (clusters), 1D (chains), 2D (sheets/layers), or mixed dimensions. This systematic approach resulted in the discovery of 9,139 low-dimensional materials, comprising 1,838 0D clusters, 1,760 1D chains, 3,057 2D sheets/layers, and 2,484 mixed-dimensional materials. Particularly noteworthy is the finding that 887 of the identified 2D materials exhibited weak interlayer interactions, indicating they are either experimentally easily exfoliable or potentially exfoliable. This provides a vast pool of candidates for new 2D materials akin to graphene.

Background & Context

Low-dimensional materials (0D, 1D, 2D) are expected to play a critical role in numerous advanced technologies—including next-generation electronics, catalysts, energy storage, and sensors—due to their unique physical and chemical properties (e.g., high surface area, quantum effects, anisotropy). However, the discovery of these materials has been highly challenging due to synthesis difficulties and the complexity of theoretical predictions. Traditional materials exploration has relied mainly on experimental trial-and-error or limited computational screening, resulting in a slow pace of discovery. The fusion of computational materials science and machine learning, as demonstrated in this study, is key to overcoming these challenges and efficiently expanding the search space to accelerate the discovery of innovative low-dimensional materials.

Strategic Significance & Outlook

The massive number of low-dimensional materials discovered in this research has the potential to complement existing materials like graphene and molybdenum disulfide (MoS2) and significantly broaden their application scope. In particular, the candidates for easily exfoliable 2D materials are expected to find diverse applications in transistors, batteries, catalysts, sensors, and composite materials. This universal computational strategy will also be applied to explore other material systems and functional materials, becoming a new standard in materials science research. The further integration of AI and computational science is anticipated to lead to the discovery of unprecedented new materials, contributing to sustainable societal development. This approach significantly accelerates virtual screening, reducing the effort and time required for laboratory synthesis.

Source: https://pubs.acs.org/doi/10.1021/acs.chemmater.5c03151?ref=PDF

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.