Prestrain Programmable 4D Printing of Nanoceramic Composites with Bioinspired Microstructure.

Four-dimensional (4D) printing enables programmable, predictable, and precise shape change of responsive materials to achieve desirable behaviors beyond conventional three-dimensional (3D) printing. However, applying 4D printing to ceramics remains challenging due to their intrinsic brittleness and inadequate stimuli-responsive ability. Here, this work proposes a conceptional combination of bioinspired microstructure design and a programmable prestrain approach for 4D printing of nanoceramics. To overcome the flexibility limitation, the bioinspired concentric cylinder structure in the struts of 3D printed lattices are replicated to develop origami nanoceramic composites with high inorganic content (95 wt%). Furthermore, 4D printing is achieved by applying a programmed prestrain to the printed lattices, enabling the desired deformation when the prestrain is released. Due to the bioinspired concentric cylinder microstructures, the printed flexible nanoceramic composites exhibit superior mechanical performance and anisotropic thermal management capability. Further, by introducing oxygen vacancies to the ceramic nanosheets, conductive nanoceramic composites are prepared with a unique sensing capability for various sensing applications. Hence, this research breaks through the limitation of ceramics in 4D printing and achieves high-performance shape morphing materials for applications under extreme conditions, such as space exploration and high-temperature systems.

Highly Robust Conductive Organo-Hydrogels with Powerful Sensing Capabilities Under Large Mechanical Stress.

The low mechanical strength of conductive hydrogels (<1 MPa) has been a significant hurdle in their practical application, as they are prone to fracturing under complex conditions, limiting their effectiveness. Here, this work fabricates a strong and tough conductive hierarchical poly(vinyl alcohol) (PEDOT:PSS/PVA) organo-hydrogel (PPS organo-hydrogel) via a facile combining strategy of self-assembly and stretch training. With PVA/PEDOT:PSS microlayers and aligned PVA/PEDOT:PSS nanofibers, PVA and PEDOT:PSS nanocrystalline domains, and semi-interpenetrating polymer networks, PPS organo-hydrogels display outstanding mechanical performances (strength: 54.8 MPa, toughness: 153.97 MJ m-3 ). Additionally, PPS organo-hydrogels also exhibit powerful sensing capabilities (gauge factor (GF): 983) due to the aligned hierarchical structures and organic liquid phase of DMSO. Notably, with the synergy of such mechanical and sensing properties, organo-hydrogels can even detect objects as light as 1 gram, despite bearing a tensile strength of ≈23 MPa. By incorporating these materials into human-machine interfaces, such as controlling artificial arms for grabbing objects and monitoring sport behaviors in soccer training, this work has unlocked a new realm of possibilities for these high-performance hierarchical organo-hydrogels. This approach to designing hierarchical structures has the potential to lead to even more high-performance hydrogels in the future.

3D printable strong and tough composite organo-hydrogels inspired by natural hierarchical composite design principles.

Fabrication of composite hydrogels can effectively enhance the mechanical and functional properties of conventional hydrogels. While ceramic reinforcement is common in many hard biological tissues, ceramic-reinforced hydrogels lack a similar natural prototype for bioinspiration. This raises a key question: How can we still attain bioinspired mechanical mechanisms in composite hydrogels without mimicking a specific composition and structure? Abstracting the hierarchical composite design principles of natural materials, this study proposes a hierarchical fabrication strategy for ceramic-reinforced organo-hydrogels, featuring (1) aligned ceramic platelets through direct-ink-write printing, (2) poly(vinyl alcohol) organo-hydrogel matrix reinforced by solution substitution, and (3) silane-treated platelet-matrix interfaces. Unit filaments are further printed into a selection of bioinspired macro-architectures, leading to high stiffness, strength, and toughness (fracture energy up to 31.1 kJ/m2), achieved through synergistic multi-scale energy dissipation. The materials also exhibit wide operation tolerance and electrical conductivity for flexible electronics in mechanically demanding conditions. Hence, this study demonstrates a model strategy that extends the fundamental design principles of natural materials to fabricate composite hydrogels with synergistic mechanical and functional enhancement.

Robust and sensitive conductive nanocomposite hydrogel with bridge cross-linking-dominated hierarchical structural design.

Conductive hydrogels have a remarkable potential for applications in soft electronics and robotics, owing to their noteworthy attributes, including electrical conductivity, stretchability, biocompatibility, etc. However, the limited strength and toughness of these hydrogels have traditionally impeded their practical implementation. Inspired by the hierarchical architecture of high-performance biological composites found in nature, we successfully fabricate a robust and sensitive conductive nanocomposite hydrogel through self-assembly-induced bridge cross-linking of MgB2 nanosheets and polyvinyl alcohol hydrogels. By combining the hierarchical lamellar microstructure with robust molecular B─O─C covalent bonds, the resulting conductive hydrogel exhibits an exceptional strength and toughness. Moreover, the hydrogel demonstrates exceptional sensitivity (response/relaxation time, 20 milliseconds; detection lower limit, ~1 Pascal) under external deformation. Such characteristics enable the conductive hydrogel to exhibit superior performance in soft sensing applications. This study introduces a high-performance conductive hydrogel and opens up exciting possibilities for the development of soft electronics.

Customizable Resilient Multifunctional Graphene Aerogels via Blend-spinning assisted Freeze Casting.

Graphene aerogels have gained considerable attention due to their unique physical properties, but their poor mechanical properties and lack of functionality have hindered their advanced applications. In this study, we propose a blend-spinning-assisted freeze-casting (BSFC) strategy to incorporate particle-modified carbon fibers into graphene aerogels for mechanical strengthening and functional enhancement. This method offers a great deal of freedom in the creation of customizable multimaterial, multiscale structural graphene aerogels. For example, we fabricated silicon carbide particle modified carbon fiber reinforced graphene (SiC/CF-GA) aerogels. The resulting aerogels display excellent properties such as being ultralightweight and highly resilient and having fatigue compression resistance (1000 cycles at 50% strain). Meanwhile, enhanced resilience inspired the effective strain-sensing capabilities of SiC/CF-GA aerogels with a sensitivity of 13.8 kPa-1. The adjustable dielectric properties due to SiC particle incorporation endow the SiC/CF-GA aerogel with a broad-band (8.0 GHz) effective electromagnetic wave attenuation performance. Besides, different particles could be incorporated into graphene aerogels via the BSFC strategy, allowing for customizable designs. Moreover, multifunctionalities were demonstrated by the modified aerogels, including noise absorption, thermal insulation, fire resistance, and waterproofing, further diversifying their practicality. Hence, the BSFC strategy provides a customized solution for fabricating modified graphene aerogels for advanced functional applications.

A note on factor normalization for deep neural network models.

Deep neural network (DNN) models often involve high-dimensional features. In most cases, these high-dimensional features can be decomposed into two parts: a low-dimensional factor and residual features with much-reduced variability and inter-feature correlation. This decomposition has several interesting theoretical implications for DNN training. Based on these implications, we develop a novel factor normalization method for better performance. The proposed method leads to a new deep learning model with two important characteristics. First, it allows factor-related feature extraction, and second, it allows for adaptive learning rates for factors and residuals. These model features improve the convergence speed on both training and testing datasets. Multiple empirical experiments are presented to demonstrate the model's superior performance.