Design, Fabrication, and Intelligent Optimisation of Electrospun Poly (ε-caprolactone) Nanofibrous Membranes for Tissue Engineering and Water Treatment Applications

Tissue engineering and water treatment are two enduring technological frontiers with far-reaching implications for human health and environmental sustainability. The dual challenges of repairing damaged tissues and addressing water pollution demand advanced, multifunctional materials that are adaptable, efficient, and environmentally responsible. Electrospinning offers control over membrane morphology, porosity, and structure, making it particularly suitable for both applications. Among available materials, electrospun poly(ε-caprolactone) (PCL) has attracted increasing attention due to its processability, biocompatibility, and biodegradability. This study aims to elucidate the parameter–structure–performance relationships of electrospun PCL-based nanofibrous materials in the contexts of tissue engineering and oily water treatment. An image-based artificial intelligence (AI) workflow was also developed to enable predictive modelling of fibre diameter, thereby improving fabrication efficiency and reducing experimental workload. Additionally, membrane modification strategies were critically reviewed to provide practical guidance for optimising design. For tissue engineering, electrospun PCL scaffolds were fabricated using polymers with varying molecular weights (MWs) under different solution concentrations and flow rates. Morphological features were assessed via scanning electron microscopy (SEM), wettability was evaluated by water contact angle (WCA) measurements, and biocompatibility was examined using L929 fibroblast cultures through MTT assays, SEM, and confocal fluorescence imaging. The MMW-25%-1.5 and HMW-20%-1.5 scaffolds exhibited optimal characteristics, including bead-free fibres (~470–490 nm), appropriate pore sizes (1.4–1.7 µm), and high porosity (>58%), which facilitated cell adhesion and proliferation. However, all scaffolds exhibited intrinsic hydrophobicity, which limited nutrient infiltration and cell adhesion. Additionally, an unexpected interaction between PCL and MTT reagents was observed, suggesting potential interference with the accuracy of the viability assay. Incorporating biomass-derived hydrophilic polymers is indicated as a strategy to enhance scaffold performance. For water treatment, sandwich-structured PCL/PMMA@PCL/PCL membranes were developed with PCL outer layers and a PMMA@PCL middle layer to enhance mechanical strength and separation efficiency. Ethanol treatment improved wettability via physical adsorption of hydroxyl groups, while cold-pressing reduced fibre diameter and increased membrane compactness through axial stretching. Compared to single-layer membranes, the multilayered structure showed improved mechanical robustness and achieved high oil rejection rates (~95%) in short-term filtration. While the flux recovery ratio (FRR) remained above 90%, it was limited by the inefficacy of deionised (DI) water rinsing in removing trapped oil. Long-term tests revealed a gradual decline in flux due to compaction and partial pore blockage, highlighting the need for improved antifouling and cleaning strategies. To address the limitations of manual fibre diameter measurement, an image-based AI workflow was developed. The DiameterJ and Semiautomated Image Measurements of Polymers (SIMPoly) programmes were compared for SEM image processing, with DiameterJ selected for its superior accuracy and batch-processing capabilities. A dataset of 144 samples was generated and used to train an artificial neural network (ANN) model using four key electrospinning parameters: molecular weight, concentration, flow rate, and tip-to-collector distance. The ANN achieved high predictive accuracy (R2 > 0.97, prediction error < 4%) and outperformed the response surface methodology (RSM) model, which showed limited generalisability (prediction error up to 28.57%). Sensitivity analysis via index of relative importance (IRI) and contour mapping identified molecular weight and solution concentration as the most influential factors, consistent with experimental observations. In summary, this study provides a comprehensive investigation into the fabrication, characterisation, and intelligent optimisation of electrospun PCL-based nanofibrous materials. It establishes a unified framework linking processing parameters, structural features, and functional performance in both tissue engineering and water treatment. The integration of macro-scale experiments with data-driven modelling offers a scalable and transferable approach to intelligent material design, paving the way for next-generation fibrous membranes with enhanced multifunctionality and environmental sustainability.