Biomimetic Mechanical Robust Cement-Resin Composites with Machine Learning- Assisted Gradient Brick-and-Mortar Structures

Traditional cement-based materials are commonly used worldwide because of their low cost and environmental adaptability, however, they the limitations of brittleness, low tensile strength, and limited toughness. Common methods engineers used to enhance cement matrices such as adding fibers or nanoparticles have resulted in only slight improvements due to the poor dispersion and inconsistent distribution of these reinforcements. Therefore, it is still a challenge to develop a material that can simultaneously offer high strength, toughness, and resistance to damage but in the same time scalable for industrial use.  Biological materials often achieve superior mechanical performance through complex and organized architectures that have evolved to withstand various stresses, however, to replicate these structures in engineering materials has been difficult due to the complexity of fabrication methods and limitations in achieving large-scale production. To this account, new study published in Advanced Materials journal and conducted by Dr. Zhangyu Wu, Dr. Hao Pan, Peng Huang, Dr. Jinhui Tang, and led by Prof. Wei She from the Jiangsu Key Laboratory of Construction Materials at School of Materials Science and Engineering in Southeast University, the researchers explored a biomimetic approach inspired by the hierarchical structures found in nacre and bamboo. They combined a machine learning-assisted design process with a scalable bottom-up fabrication strategy and created a gradient brick-and-mortar structure that significantly enhances mechanical properties of cement-based marterials.

The authors began the fabrication process with the creation of damp cement particle aggregates and produced by mixing aluminate cement with polyvinyl alcohol and deionized water. This mixture was then subjected to high-speed shearing to create plastic sheets which were crushed into varying particle sizes. They coated these chippings with thermosetting resin and stacked in a specific order to form a gradient brick-and-mortar structure, inspired by natural materials like nacre and bamboo. The researchers then hot-pressed the samples at 150°C and 100 MPa to finalize the formation of the composite, creating both homogeneous and heterogeneous samples for comparison. Afterward, the team conducted a series of quasi-static flexural tests to assess the mechanical properties of these composites which showed that the gradient composite outperformed both homogeneous and heterogeneous composites in terms of strength and toughness. Specifically, they found that the gradient composite has a flexural strength up to 70.07 MPa, a 192% increase compared to its heterogeneous counterpart. Additionally, the toughness of the gradient composite was found to be 4765% higher than that of the heterogeneous composite which demonstrated its ability to withstand significant deformation before failure. That excellent performance of the gradient structure was thought to be because of the regional densification of the matrix and the crack-tip shielding effects which was facilitated by the well-organized brick-and-mortar configuration. Moreover, the researchers evaluated how the material would respond to both low- and high-velocity impacts using drop-hammer and ballistic impact tests and found the gradient composite has a peak force of approximately 22,951.8 N and an energy absorption capability of 17.8 J, show higher performance than pure cement, homogeneous composites, and even some natural rocks and alloys. Additionally, under high-velocity impact conditions, the gradient composite showed remarkable resilience, which sustained minor surface cratering while maintaining its structural integrity while pure cement fractured entirely. The authors’ findings were also confirmed through high-speed camera imaging and finite element simulations, which illustrated how the composite’s unique structure deflected cracks and absorbed impact energy efficiently. Furthermore, the researchers investigated the microstructural characteristics of the composites to better understand the mechanisms behind their enhanced mechanical properties using scanning electron microscopy and energy dispersive spectroscopy (EDS) analyses which demonstrated a well-defined gradient brick-and-mortar structure with the denser bottom layers that provided strength and porous top layers that gave flexibility. The EDS mapping confirmed the successful integration of cement and resin in the composite with distinct regions of aluminum, calcium, and carbon corresponding to the different phases. Microcracks observed in the resin “mortar” phase, along with polymer bridges over cracks in the cement “brick” phase, highlighted the material’s capacity for crack deflection and inhibition, contributing to its enhanced toughness.

Dr. Zhangyu Wu and colleagues optimized the design of the composite further using machine learning model that integrated a backpropagation neural network (BPNN) and a genetic algorithm (GA). The model allowed the team to predict and optimize the structural parameters of the composite, such as particle size and resin content to achieve maximum flexural strength. After training the BPNN on a dataset of 336 samples and running the GA for 100 iterations, the new model identified the optimal configuration for the gradient structure which was then experimentally validated. The optimized gradient composite demonstrated much higher strength and toughness than the unoptimized samples with test results closely matching the predictions of the machine learning model. In conclusion, The scholars in Southeast University successfully developed a cement-resin composite with significantly enhanced mechanical properties using natural materials and advanced machine learning techniques and by this addressed the limitations of traditional cement-based materials in particular their brittleness and poor toughness. The new research demonstrated indeed the power of biomimetic design and highlighted the potential of machine learning to optimize complex material structures efficiently, techniques that can be used in the future to develop and design more sophisticated advanced composites tailored for specific engineering applications. According to the authors, the demonstrated scalability of the fabrication process also addresses one of the primary challenges in the field of biomimetic materials of scaling up production but in the same time maintaining complex and hierarchical structures. This ability to mass-produce optimized materials with superior performance characteristics has significant potential in industries such as construction, aerospace, and defense, where materials must endure extreme conditions without failure.