Despite the presence of nanodiamond (ND) agglomeration in composites, this critical factor has been unnoticed in prior models predicting the tensile modulus of ND-filled systems. In the present study, the properties of agglomerates and their surrounding interphase are incorporated to state a predictive expression for the Young’s modulus in the ND-reinforced samples. The operative volume share of reinforcement (ϕeff) is defined by agglomerate radius (Ragg), ND size, and the interphase thickness (t). Notably, ϕeff reaches a maximum value of 0.55 at ND radius of 5 nm (indicating no agglomeration), but significantly decreases to 0.015 at Ragg = 50 nm, indicating that larger agglomerates cause the least reinforcement. The nanocomposite modulus exhibits a 420% enhancement at Ragg = 10 nm; however, further increases in agglomerate size markedly reduce stiffness. Similarly, a 300% increase in modulus is attained at t = 20 nm, whereas thinner interphases lead to diminished reinforcement. A stiffer interphase also improves mechanical performance; 215% enhancement in the stiffness is seen at an interphase modulus of 30 GPa. In contrast, a lower interphase modulus of 5 GPa induces only a marginal 35% improvement of nanocomposite modulus. The proposed model substantially overestimates the modulus when assuming well-dispersed NDs, but accurately predicts the stiffness of real composites containing agglomerated NDs.
Hybrid 3D braided composites composed of high-strength T800 and high-modulus M55 carbon fibers must simultaneously coordinate stiffness utilization, strain mismatch, interface-mediated load transfer, and fiber integrity during surface activation. In this study, T800/M55 hybrid 3D five-directional braided composites with an epoxy vinyl ester resin matrix were investigated by integrating periodic unit-cell screening, fiber-specific plasma treatment, multiscale surface characterization, single-filament breaking-load and Weibull analyses, unidirectional short-beam shear (SBS) screening, and multi-mode mechanical validation. Model-guided screening showed that the M55 axial yarn/T800 braiding yarn configuration (H3) provided a favorable balance between axial stiffness gain and stress localization. Moderate T800-O2-4 min and M55-N2-3 min treatments each retained approximately 91% of the untreated mean single-filament breaking load, whereas prolonged treatment reduced load retention and/or Weibull statistical uniformity. These two controlled activation windows also maximized the SBS responses of the corresponding unidirectional composites. After these treatment windows were applied to H3, the mean values of SBS strength, flexural strength, tensile modulus, failure strain, and compressive strength increased, whereas the mean flexural modulus decreased. Fracture morphology progressively changed from clean fiber pull-out and interfacial debonding to greater resin coverage, matrix tearing, and mixed interfacial/cohesive fracture. These results reveal the competition between interfacial activation benefits and treatment-induced fiber degradation and establish an architecture–activation–load-transfer correlation in hybrid textile composites.
Polypyrrole-NiFe2O4-CaCu3Ti4O12 (PNC) nanocomposites are prepared via in-situ chemical oxidative polymerization. In this process, the weight ratio of pyrrole monomer and NiFe2O4 (NFO) is fixed to 1:1 whereas the concentration of CaCu3Ti4O12 (CCTO) is varied from 0 to 1 by weight fraction in the step size of 0.25 to study the influence of dielectric filler incorporation on electromagnetic shielding effectiveness of polymer-based nanocomposites in X-band. XRD and FTIR confirms the successful formation of a multi-phase nanocomposite structure. FESEM imaging reveals increased porosity and surface roughness with increasing CCTO content. Further, microscopic studies confirm the formation of cage-like structures where the nanofillers are attached with the polymeric chain. The nanocomposites exhibit ohmic electrical behaviour with DC conductivity of ∼ 0.2 S cm-1, and ferrimagnetic behaviour with a coercivity ∼ 330 Oe for PNC1. The electromagnetic interference (EMI) shielding observations emphasis on enhanced absorption tendency of the nanocomposites with the inclusion of CCTO. The highest shielding effectiveness (SE) value is achieved for PNC1 (1 wt. % of both nanofillers) ∼ 38 dB with absorption contribution ∼ 32 dB along with absorption efficiency ∼ 99.7%. Microwave dielectric analysis shows a significant increase in relative permittivity (εr) ∼ 65 and attenuation coefficient (α) ∼ 800 Np/m with CCTO loading. Reflection loss (RL) studies further confirm the microwave absorption capability of PNC1, which is approximately −10 dB. The synergistic contribution of dielectric and magnetic losses in PNC nanocomposites corresponds to superior EMI shielding performance, which could be useful in electronic, defence, communication, and aerospace applications.
Carbon fiber reinforced polymer (CFRP) composites are crucial to manufacture the high load-bearing structures with lightweight, yet their mechanical performance depends on interacting design variables, such as fiber fraction, shape, and length. These factors are usually determined through expensive testing and simulations that require high computational cost and even induce an ultrahigh challenge.In this work, we introduce a coupled framework of machine learning (ML) and finite element method (FEM) to both predict and inversely design the mechanical properties under transverse and longitudinal tensile loading. FEM was used to generate a broad design-response space, and generative adversarial networks (GAN) were employed to augment the dataset to improve coverage, and mutual-information-based selection identified the most informative features. The trained ML regressor achieved high accuracy (R2 ≥ 0.9912) and showed strong agreement with independent experimental data. Model interpretability via SHapley Additive exPlanations (SHAP) reveals that Young’s modulus is governed primarily by the volume fraction, followed by shape and length of fiber, providing the design-relevant insights rather than black-box fits. In addition to forward mechanical property prediction, Bayesian optimization (BO) is coupled with the trained machine learning model to implement target-driven inverse microstructural design. The resulting optimal microstructural candidates were further validated via FEM simulation. The findings indicate that the FEM-GAN-ML-BO framework is capable of efficiently forecasting, analyzing, and inversely designing CFRP composites. This framework minimizes the necessity for extensive experimental trials and large-scale numerical simulations, thereby aiding the advancement of high-performance composite materials for engineering purposes.
Chitosan-derived carbon aerogels have attracted significant attention in the field of electromagnetic wave (EMW) absorption due to their lightweight characteristic, simple preparation process and three-dimensional (3D) porous structure. However, the traditional manufacturing process still encounters numerous challenges in fabricating chitosan-derived carbon aerogels as multifunctional EMW absorbers. This highlights the necessity of developing simple and structurally sophisticated strategies. In this work, chitosan-derived carbon/cobalt (CDC/Co) composite aerogels with the low bulk density (42−57.2 mg/cm3) were prepared through a three-step method of physical crosslinking, freeze casting and high-temperature carbonization. The research results showed that the obtained binary composite aerogels had a unique 3D porous network structure, and numerous cobalt particles were grown in situ and distributed on the surface of the carbon framework. Furthermore, the abundant CDC/Co heterogeneous interfaces could induce the interfacial polarization under the alternating electromagnetic fields, thereby endowing CDC/Co composite aerogel with excellent EMW absorption performance. Significantly, the CDC/Co composite aerogel with the addition amount of cobalt salt of 0.4 mmol and a filling ratio of 10 wt.% achieved the minimum reflection loss of -48.20 dB at a matching thickness of 3.34 mm and the maximum effective absorption bandwidth of 6.88 GHz under a thin thickness of 2.26 mm. Additionally, the CDC/Co composite aerogel also exhibited superior radar stealth and thermal insulation properties. Therefore, this study provides a new idea for the development of biomass chitosan-derived carbon-based composite aerogels as multifunctional EMW absorbers.
In this study, two invariant-based failure criteria are proposed to predict failure of a unidirectional non-crimp fabric (UD-NCF) carbon fiber/epoxy composite. Three different failure modes, namely matrix failure, fiber tensile failure, and fiber compressive failure, are considered. The matrix failure functions are based on linear and quadratic transversely isotropic stress-invariants, while fiber failure is evaluated with a maximum stress criterion for tension and an adapted kinking criterion for compression. The main advantages of the invariant-based criteria are fewer required experiments for calibration and minimal computational effort in comparison with fracture plane-based criteria. Moreover, the chosen mathematical formulation ensures a non-negative Hessian and decoupled tensile and compressive behaviour. Furthermore, the strain rate dependency of the investigated UD-NCF composite is incorporated using empirical relations defined in previous work by the authors. For validation purposes, off-axis tests are performed at quasi-static and high strain rates. Overall, the observed agreement with experimental data presented here as well as from literature is excellent, highlighting the feasibility of the proposed criteria for design of UD-NCF composite structures.
Thermal comfort is increasingly challenged by heat waves and abrupt temperature fluctuations, requiring materials that dissipate heat while buffering thermal shocks. Existing combinations of radiative cooling and phase change systems are often achieved by stacking or mixing components, leaving structural design and performance regulation largely independent, which limits the mutual reinforcement of steady cooling and transient thermal regulation. Herein, we employ coaxial electrospinning to construct a core-shell fibrous film composed of polyethylene glycol@poly (vinylidene fluoride-co-hexafluoropropylene)/cellulose nanocrystals (PEG@PVDF-HFP/CNCs) for the synergistic regulation of radiative cooling and phase change performance. The shell achieves 94.91% solar reflectivity (0.3-2.5 μm) and 97.56% infrared emissivity (8-13 μm), while the core provides high thermal shock resistance with a phase change enthalpy of 47.3 J g−1. Crucially, CNCs enhance radiative cooling by optimizing structural scale and optical parameters, while simultaneously improving phase change thermal regulation by inducing PEG crystallization to enhance latent heat storage, achieving synergistic enhancement of both functions within the same material system. Outdoor measurements demonstrate a steady-state temperature reduction of 10.5°C together with effective buffering against abrupt ambient temperature fluctuations. This work achieves synergistic optimization of radiative cooling and phase change performance within one system, providing a novel pathway for thermal comfort regulation under dynamic environments.
With the increasing severity of electromagnetic pollution and the limitations of conventional shielding technologies, microwave absorbing materials have attracted significant attention for their efficient conversion of electromagnetic energy into heat. In this work, boron/nitrogen-doped hollow carbon nanospheres (B,N-CNs) were prepared by oxidation polymerization. Subsequently, the B,N-CNs were loaded on MXene and incorporated into a polyurethane (PU) hydrogel matrix to fabricate a B,N-CNs/MXene/PU gel composite absorber. The synergistic combination of the high electrical conductivity of MXene and the hollow structure of B,N-CNs generates abundant heterogeneous interfaces, facilitating multiple polarization processes and enhanced electromagnetic wave attenuation. Moreover, B,N co-doping introduces abundant defects and heteroatomic sites, which strengthens defect-induced dielectric loss while optimizing impedance matching. Benefiting from this multi-component and multi-scale structural design, the B,N-CNs/MXene/PU gel exhibits improved microwave absorption performance, achieving a minimum reflection loss (RLmin) of −41.4 dB at 8.8 GHz, along with an effective absorption bandwidth (EAB) of 2.5 GHz at a thickness of 2.5 mm. Furthermore, the incorporation of B,N-CNs significantly reinforces tensile properties of the composite, with the elongation at break increasing from 790% for MXene/PU gel to 1125% for B,N-CNs/MXene/PU gel.
Broadband and strong absorption properties are desirable characteristics for electromagnetic wave (EMW) absorbing materials; however, simultaneously achieving both broad bandwidth and strong absorption remains a major challenge. Here, we propose a multiscale structural design strategy that integrates a microscale segregated network with a macroscale non-gradient multilayer structure. Typically, negatively charged MXene@Ni nanosheets were electrostatically assembled onto the surface of polyether ether ketone (PEEK) and subsequently hot-pressed to form the single-layer PEEK/MXene@Ni (PMN) composites with microscale segregated conductive network. At 5 wt% filler loading, the PMN composite achieved a minimum reflection loss (RLmin) of −43.77 dB and a maximum effective absorption bandwidth (EAB) of 3.84 GHz due to the interfacial polarization and multiple reflection effects in the segregated network. To further broaden the EAB, an impedance matching-guided multilayer engineering was employed to develop the non-gradient multi-layer structure at the macroscale. Accordingly, the four-layer non-gradient composite (NG-4) exhibited an RLmin of −68.23 dB and a maximum EAB as high as 10.56 GHz. This exceptional performance is mainly attributed to the synergistic effect of impedance matching for EMW penetration and cascade dissipation through the multi-layer structure. Therefore, this multiscale design, ranging from microscale segregated networks to macroscale multilayer structures, provides a promising new pathway toward broadband and high-efficiency EMW absorbing materials.
Recent advancements in deep learning offer considerable potential for automated composite damage recognition. However, limited annotated datasets and complex material microstructures challenge pixel-level characterization of local damage mechanisms, while existing approaches provide limited integration of segmentation outputs with quantitative damage assessment. A deep learning-based convolutional neural network was integrated with digital imaging and automated crack counting algorithms to characterize damage in unidirectional non-crimp fabric glass fiber-reinforced reactive thermoplastic cross-ply laminates. Microscopic images of damage-containing material samples obtained in previous work were preprocessed and annotated with six feature-related classes for model training and testing. A U-Net architecture incorporating skip connections, dropout, and a weighted Dice loss function was employed to segment material features and damage. Results showed that smaller batch sizes produced the lowest weighted Dice loss and the highest Dice coefficients. Notably, the Dice coefficient improved significantly from ∼0.736 with equal class weights to ∼0.840 with optimized class weights, representing a 14.1% increase in the prediction accuracy of 90° tow cracks (the dominant damage mechanism). A controlled factorial ablation study further evaluated key model components across eight configurations and three random seeds. Removing skip connections from the full model reduced the mean Dice coefficient by approximately 51.6%, demonstrating their importance for multi-class segmentation. The trained model demonstrated a strong capability to identify damage initiation, progression, and interaction, and was integrated with automated crack counting algorithms for quantitative damage assessment. The resulting damage metrics could support analytical damage-mechanics models for predicting stiffness degradation and informing structural health monitoring and maintenance planning.
Epoxy fiber-reinforced polymers (FRPs) are extensively utilized in high-performance structural parts owing to their outstanding strength and rigidity. Nevertheless, their inherent brittleness and irreparability significantly shorten their service life, especially in demanding environments. Herein, a novel class of epoxy FRPs featuring high strength, enhanced toughness, and self-healing was presented via a dual-network curing design incorporating both rigid and flexible curing agents. Precise tuning of the network structural parameter R yielded well-balanced mechanical attributes, including tensile strength surpassing 400 MPa, tearing energy up to 430 kJ/m2, and healing efficiencies exceeding 90% across multiple damage-healing cycles. Dynamic boronic ester linkages embedded in the matrix enabled molecular rearrangements at elevated temperatures while preserving the thermoset framework. Mechanical evaluations and fracture morphology analyses indicated that an optimal R window (115-250) synergistically enhanced energy dissipation, interfacial bonding, and fiber-matrix stress transfer. This work establishes a design paradigm that harmonizes strength, toughness, and reparability, facilitating the development of durable and maintainable thermosetting composites for advanced structural applications.
Surfaces of renewable energy infrastructure like wind turbine blades and photovoltaic panels are facing severe challenges under harsh service environments such as icing, sandstorms and contaminations. Protective outer coatings with low ice adhesion, superior mechanical durability, and high optical transmittance are urgently needed. Herein, we propose a multiscale puzzle complementary strategy to design transparent and mechanically robust self-lubricating surface (TRSS). The self-lubricating constituent (SLC) that synthesized via a molecular-scale puzzle assembly effect endows the surface with ultralow surface energy to achieve self-lubrication and low ice adhesion. Meanwhile, a macroscale interpenetrating polymer network puzzle via integrating reinforcing constituent and SLC drastically improves the composite’s bulk mechanical integrity. The as-prepared TRSS exhibits excellent mechanical robustness, a low friction coefficient, high optical transparency, and outstanding dynamic anti-icing performance. Its remarkable light transmittance minimizes power generation loss when coated on photovoltaic panels, validating its promising applicability in photovoltaic industry. Systematic characterizations reveal that TRSS possesses prominent anti-icing and deicing efficacy with an ice adhesion strength of ∼30 kPa, and exceptional wear resistance where only ∼80 μm thickness loss is observed after 800 cycles of Taber abrasion. Collectively, these comprehensive functional merits establish TRSS as a dependable protective candidate for both aircrafts and renewable energy facilities, possessing great practical significance and broad engineering deployment prospects.
High-temperature electromagnetic wave absorbing materials are crucial for electromagnetic protection and military stealth. Ceramizable composites are promising candidates for integrating microwave absorption, load-bearing capacity, and thermal protection in multi-field coupling environments; however, their high-temperature electromagnetic response remains poorly understood. This study presents the first systematic investigation of the effects of pretreatment temperature and in situ test temperature on electromagnetic wave absorption of quartz fabric-reinforced phenolic-based ceramizable composites. Experimental results show that increasing pretreatment temperature to 800 °C promotes pore development and heterogeneous interfaces, raising the average X-band reflection loss (RL) from −2 dB to −8 dB. Complete ceramization at 1100 °C forms a stable ZrO2/Al2O3 ceramic skeleton, achieving a minimum RL of −10 dB. For fully ceramized specimens, absorption first improves with test temperature up to 400 °C (RL≈−12 dB) due to activated polarization, then declines slightly at 600 °C owing to excessive thermal perturbation. To enable performance prediction, a thermal–chemical–electromagnetic multi-physics coupled simulation framework is established, based on thermogravimetric analysis and reaction kinetics. By mapping reaction progress to complex permittivity, the model successfully predicts RL under various service conditions and reveals parametric sensitivities of thickness, oxygen concentration, and holding time. This work provides a unified experimental and theoretical basis for the design of high-temperature microwave absorption materials.
Continuous fibre-reinforced nylon produced by fused filament fabrication (FFF) offers great potential for lightweight structural applications; however, process-induced defects such as porosity, fibre misalignment and inter-bead discontinuities can significantly reduce compressive performance. This study investigates the role of these microstructural features in governing damage evolution and collapse in continuous carbon fibre-reinforced (CFRP) and glass fibre-reinforced (GFRP) nylon under uniaxial compression using in-situ synchrotron X-ray computed tomography (XCT). XCT scans recorded before and after compression are analysed using a deep learning-based segmentation framework to quantify phase fractions, fibre orientation and void morphology. CFRP exhibits higher void content (11%–12%) and lower fibre volume fractions (30%–33%) than GFRP. Approximately 70% and 50% of the fibres are oriented within 2.5° of the extrusion direction in CFRP and GFRP, respectively. For both CFRP samples and one GFRP sample, kinking was observed out-of-plane of the printed layers, accompanied by fibre fracture and delamination. In the other case, the GFRP exhibited more distributed deformation characterised by fibre reorientation, bending instability and interlaminar damage. This was ascribed to a reduction in lateral constraint arising from interlaminar voids, but the lower degree of fibre alignment for that sample may also contribute. These results show how compressive failure in continuous-fibre FFF composites is controlled by the coupled interactions between fibre architecture, void morphology and nonuniform local stress field, providing a basis for microstructure-informed process optimisation and improved structural performance.
