Billions of Tiny Seeds: Machine Learning Unlocks Scalable Growth of Carbon Nanotubes

Carbon nanotube growth via nanoparticles

In the race to develop ultra-compact, high-performance electronics, nanomaterials like carbon nanotubes (CNTs) have emerged as promising candidates to overcome current limitations in heat dissipation and miniaturization. Now, researchers from the University of Pittsburgh and Rutgers University are combining cutting-edge microscopy with machine learning to unravel the mystery of how to grow these materials in a more controlled, scalable way.

From Grass to Graphene: Cultivating CNT Forests

Led by Dr. Mostafa Bedewy and Dr. Ahmed Aziz Ezzat, the research team received a $549,947 grant from the U.S. National Science Foundation to investigate the dynamics of carbon nanotube growth at the nanoscale. Their work focuses on alumina-supported iron nanoparticles—described metaphorically as “billions of seeds”—spread over tiny plots just one centimeter square.

“These nanoparticles are like tiny seeds, and some of them sprout into nanotubes while others don’t,” Bedewy explains. “Our goal is to discover why that happens and to learn how to guide this process predictively.”

The Role of Machine Learning and E-TEM

To study these processes at atomic resolution, the researchers utilize an advanced imaging technique known as environmental transmission electron microscopy (E-TEM). E-TEM captures detailed images of nanoparticles in action — but traditionally, analyzing these images has been a slow, manual effort.

Now, with the help of machine learning algorithms, the team can process hundreds of high-resolution images per second. These algorithms track the spatio-temporal evolution of nanoparticles and help predict how they will behave during chemical vapor deposition (CVD), the process used to synthesize carbon nanotubes.

Why Carbon Nanotubes Matter

Carbon nanotubes are tiny tube-shaped structures composed of carbon atoms arranged in a hexagonal lattice. They are incredibly strong — over 100 times stronger than steel — and conduct electricity far better than copper. Because of these properties, CNTs are being explored for use in everything from next-generation computer chips and energy storage systems to aerospace components and water purification membranes.

Recent progress even includes carbon-nanotube-based Tensor Processing Units (TPUs) developed in China, which could potentially replace silicon in future AI accelerators due to their energy efficiency and thermal conductivity.

Predicting the Unpredictable

This project stands out not just for its focus on material properties, but also for its data science rigor. “We’re not just trying to grow CNTs,” says co-investigator Ezzat. “We’re aiming to build a predictive simulation tool that can analyze complex atomic-scale behavior from E-TEM data in real-time.”

If successful, their approach could revolutionize the way engineers approach nanomanufacturing — turning trial-and-error experiments into precise, predictable fabrication strategies.

A Seedbed for the Future

By marrying machine learning and in situ microscopy, this research highlights how nanotechnology and AI can converge to reshape the electronics landscape. As we continue to miniaturize and densify our devices, learning how to reliably “plant and grow” carbon nanotubes could be the key to unlocking the next era of energy-efficient, ultra-fast computing hardware.

🔗 Original article published by Interesting Engineering: https://interestingengineering.com/innovation/billion-tiny-seeds-grow-carbon-nanotubes

Sponsored by PWmat (Lonxun Quantum) – a leading developer of GPU-accelerated materials simulation software for cutting-edge quantum, energy, and semiconductor research. Learn more about our solutions at: https://www.pwmat.com/en

📞 Phone: +86 400-618-6006
📧 Email: support@pwmat.com

Comments

Post a Comment

Popular posts from this blog

Quantum Chemistry Meets AI: A New Era for Molecular Machine Learning

Image
In a powerful leap forward for computational chemistry and materials design, researchers from Carnegie Mellon University have developed a new kind of molecular machine learning model—one that goes beyond simplistic molecular graphs and integrates fundamental principles of quantum chemistry. Their work, published in Nature Machine Intelligence , could radically enhance our ability to predict chemical behavior, optimize catalysts, and discover new drugs, all while using less data and gaining more scientific insight. The Problem: Traditional Molecular Representations Fall Short Molecular machine learning (ML) models underpin key processes in drug discovery, materials science, and catalysis. However, most existing models use simplified representations such as SMILES strings, 2D graphs, or coordinate matrices. These techniques often lack the quantum-chemical information that governs how molecules truly behave—particularly electron interactions that define reactivity, stability, and...
Read more

MIT’s React-OT: The AI Model Revolutionizing Chemical Reaction Design

Image
MIT’s React-OT: The AI Model Revolutionizing Chemical Reaction Design In a major leap for materials science and computational chemistry, researchers at MIT have unveiled a new AI-driven model called React-OT , capable of predicting the “point of no return” in chemical reactions — also known as the transition state — with remarkable speed and accuracy. Published in Phys.org on April 23, 2025, this innovation could revolutionize how scientists design sustainable chemical processes and novel materials. Why it matters: Understanding and predicting transition states is vital for designing efficient reactions in pharmaceuticals, polymers, and fuels. Traditional methods based on quantum chemistry are computationally expensive — sometimes taking days for a single calculation. Enter React-OT — a machine-learning model trained on over 9,000 reactions that slashes prediction time to under a second. The model’s secret? It begins with a smart est...
Read more

OMol25: A Record-Breaking Dataset Set to Revolutionize AI in Computational Chemistry

Image
Published on: Quantum Server Networks | Date: May 2025 In an era where artificial intelligence is reshaping every scientific frontier, a groundbreaking resource named Open Molecules 2025 (OMol25) has just been launched. This record-breaking dataset is poised to radically transform the way researchers model and simulate molecular interactions, accelerating innovation in materials science, biochemistry, and clean energy. Berkeley Lab News Center reports that the dataset, jointly developed by Meta’s Fundamental AI Research (FAIR) lab and the U.S. Department of Energy’s Lawrence Berkeley National Laboratory, encompasses over 100 million high-fidelity 3D molecular snapshots . These were generated using Density Functional Theory (DFT), a quantum mechanical modeling method renowned for its precision but traditionally limited by massive computational demands. What Makes OMol25 So Important? DFT simulations have long been a gold standard for predicting how atoms behave, includ...
Read more