Redefining Strength and Lightness: Carbon Nanolattices Optimized Through Bayesian Design

The development of nanoarchitected materials has revolutionized materials science by enabling the creation of structures that achieve remarkable mechanical properties, such as exceptional specific strength, stiffness, and energy absorption. These materials are meticulously designed at the nanoscale to harness unique mechanical behaviors not found in conventional bulk materials, making them ideal for applications in aerospace, defense, biomedical devices, and other cutting-edge technologies. Despite their promise, these materials face significant challenges that limit their practical adoption. Chief among these are stress concentration at junctions, scalability in manufacturing, and suboptimal material utilization, which collectively undermine their overall performance and hinder their integration into large-scale engineering systems. Traditionally, nanoarchitected materials are built using uniform topologies such as lattices, honeycombs, and gyroids. While these designs have demonstrated some level of success, they are inherently constrained by poor stress distribution and nodal failure, where stress concentrations at the junctions cause premature material breakdown. Moreover, achieving a balance between mechanical strength and weight—critical for applications such as lightweighting in aerospace—remains an elusive goal. Efforts to improve material performance through topology optimization have often relied on manual or conventional design approaches, which are unable to explore the vast and complex design space effectively. Another pressing issue lies in the scalability of production. Current nanoscale fabrication methods, such as two-photon polymerization (2PP), offer high resolution but are slow and labor-intensive, limiting the size of produced materials. This bottleneck prevents these promising materials from being utilized in real-world, macroscopic applications. Furthermore, the intrinsic material properties of the nanostructures themselves have yet to be fully optimized. For instance, while pyrolyzed carbon is known for its high specific strength, its performance depends heavily on the atomic structure and purity, which vary with nanoscale dimensions and manufacturing conditions. These unresolved issues create an urgent need for innovative strategies that can overcome the limitations of both design and production.

New research paper published in Advanced Materials Journal and conducted by international collaborators led by Professor Tobin Filleter from the University of Toronto and Professor Seunghwa Ryu from Korea Advanced Institute of Science and Technology (KAIST) addressed these challenges comprehensively. They recognized that the integration of advanced computational techniques, such as machine learning, with cutting-edge fabrication processes could redefine the potential of nanoarchitected materials. By employing multi-objective Bayesian optimization (MBO), the team sought to automate and enhance the design of nanolattice geometries, focusing on maximizing mechanical performance while minimizing density.   The researchers used finite element analysis to predict and optimize the performance of different lattice geometries. Starting with randomly generated initial designs, the algorithm iteratively improved the structures by balancing objectives such as stiffness, strength, and density. These optimized designs were then fabricated using two-photon polymerization (2PP), a nanoscale additive manufacturing technique that enabled the creation of highly precise polymeric lattices, which were later pyrolyzed to convert them into carbon structures.

The authors found exceptional improvement in mechanical properties achieved through the MBO-generated designs. The researchers compared the optimized nanolattices to traditional ones with uniform struts and found that the new designs exhibited up to 68% higher Young’s modulus and a remarkable 118% increase in strength at comparable densities. These enhancements were attributed to the non-intuitive redistribution of material by the MBO algorithm, which reduced stress concentrations at junctions and ensured a more uniform stress distribution throughout the structure. Scanning electron microscopy revealed that the fabricated lattices closely matched the complex geometries predicted by the algorithm, confirming the high fidelity of the manufacturing process. Further experiments focused on the nanoscale properties of the pyrolyzed carbon used to construct the lattices. By reducing strut diameters from 600 nanometers to 300 nanometers, the team observed dramatic increases in both stiffness (75%) and strength (79%). Raman spectroscopy and X-ray photoelectron spectroscopy provided insights into the atomic composition of the material, revealing a significant increase in sp²-bonded aromatic carbon in the thinner struts. This improvement was tied to the pyrolysis process, which introduced a radial gradient in atomic structure. Thinner struts exhibited a higher purity of sp² carbon and lower oxygen impurities, thanks to accelerated desorptive mass flux during pyrolysis. This gradient was further confirmed through electron energy loss spectroscopy, which showed a consistent structure across the strut cross-section in smaller diameters, enhancing mechanical properties.

To assess the scalability of their approach, the researchers fabricated macroscopic nanolattices composed of millions of unit cells using a multi-focus 2PP system. These lattices spanned millimeter scales yet retained submicron precision. One experiment demonstrated their extraordinary strength-to-weight ratio by balancing an entire lattice on a soap bubble while simultaneously supporting more than one million times its weight. Mechanical testing through uniaxial compression revealed that the optimized designs distributed stress across all lattice layers, preventing localized failure and ensuring the entire structure contributed to load-bearing. This contrasted sharply with traditional designs, which exhibited layer-by-layer failure under similar conditions. The findings highlight the synergy between advanced computational optimization, precise nanoscale fabrication, and material science. The combination of MBO-designed geometries and high-purity pyrolyzed carbon resulted in a nanolattice material that achieved a specific strength of 2.03 MPa·m³/kg—comparable to carbon steel at densities akin to Styrofoam. This study represents a transformative step in the field of nanoarchitected materials, addressing long-standing challenges and unlocking new possibilities for lightweight, high-strength structures. The researchers demonstrated that combining multi-objective Bayesian optimization with advanced nanoscale manufacturing can not only optimize mechanical performance but also fundamentally reimagine how materials are designed and used. By achieving a specific strength comparable to carbon steel while maintaining the density of materials like Styrofoam, this research establishes a new benchmark for structural materials, challenging the traditional trade-off between strength and weight.

The new study is significant because it provided a proof-of-concept for using computational tools to navigate complex design spaces, enabling the creation of structures that were previously unimaginable. These findings also highlight the critical role of material properties at the nanoscale, demonstrating how subtle changes in atomic composition and strut geometry can lead to significant improvements in mechanical performance. This insight underscores the importance of tailoring both the structural design and material chemistry to achieve unprecedented outcomes. From an application standpoint, the implications of this study are vast. In aerospace, these ultralight yet strong materials can reduce fuel consumption, increase payload capacities, and improve overall efficiency. In defense, their exceptional energy absorption properties make them ideal for protective gear and impact-resistant components. The biomedical field could also benefit greatly, with the potential for creating scaffolds for tissue engineering or lightweight implants that mimic the properties of natural bone. Additionally, their scalability, demonstrated by the production of millimeter-scale lattices with nanometer precision, suggests that these materials could be adopted for industrial-scale manufacturing, opening doors to entirely new engineering solutions. This work also emphasizes the environmental and economic benefits of material efficiency. By minimizing the density of structural materials without sacrificing performance, the study offers pathways to reduce raw material usage, energy consumption, and waste. Such advancements could play a critical role in sustainable development across industries.